{
  "scope": "Complete artifact validation plus eight authored native cached greedy fixtures; not a broad quality or speed benchmark.",
  "bits": 6,
  "backend": "Linux MLX CUDA 12 on NVIDIA B200",
  "mac_acceptance": "pending",
  "files": [
    {
      "file": "model-00001-of-00006.safetensors",
      "bytes": 5224773925,
      "sha256": "5921012d0768006242d922c058305f4690eb7776c94acf0b8c1a5d1d7a65f488"
    },
    {
      "file": "model-00002-of-00006.safetensors",
      "bytes": 5225058031,
      "sha256": "791cd63088c103b46cc4940a98be89c4bc3b7b40d31ebc030a3756d8587204b4"
    },
    {
      "file": "model-00003-of-00006.safetensors",
      "bytes": 5250337810,
      "sha256": "628d54adbb7ae77d03819eea223d3656dd07c6ba35847e0c3e09d7ec89b5c9f7"
    },
    {
      "file": "model-00004-of-00006.safetensors",
      "bytes": 5250337542,
      "sha256": "d190c36462c12b3b46a90515d87684e252e1ac509a21c4cf8ac33233fb2b5f7d"
    },
    {
      "file": "model-00005-of-00006.safetensors",
      "bytes": 5219774304,
      "sha256": "32e7a94b204ba5c328f86dd1e3bc7f572b20c197f5ca096fb62382dac41c0075"
    },
    {
      "file": "model-00006-of-00006.safetensors",
      "bytes": 1998341507,
      "sha256": "11c428fc2c54edfda8c8be74b9f0ba29ad927a1c8585e78cd35288ff32d8af6c"
    }
  ],
  "tensor_checks": [
    {
      "name": "language_model.model.layers.7.self_attn.v_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.7.self_attn.v_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.self_attn.v_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.self_attn.q_proj.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.self_attn.q_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.self_attn.o_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.self_attn.o_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.self_attn.k_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.7.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.7.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.self_attn.k_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.6.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.6.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.6.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.6.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.6.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.self_attn.q_proj.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.6.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.6.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.6.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.6.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.5.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.5.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.5.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.0.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.0.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.1.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.4.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.self_attn.k_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.2.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.3.self_attn.k_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.4.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.2.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.2.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.1.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.1.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.1.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.2.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.0.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.4.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.2.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.self_attn.k_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.6.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.1.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.5.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.self_attn.o_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.2.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.5.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.7.self_attn.o_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.2.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.3.self_attn.q_proj.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.0.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.1.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.0.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.1.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.0.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.self_attn.k_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.5.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.embed_tokens.scales",
      "shape": [
        248320,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.1.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.6.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.embed_tokens.biases",
      "shape": [
        248320,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.4.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.2.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.4.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.0.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.2.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.0.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.1.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.0.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.6.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.0.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.0.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.6.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.6.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.self_attn.o_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.self_attn.q_proj.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.embed_tokens.weight",
      "shape": [
        248320,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.0.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.1.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.2.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.self_attn.v_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.0.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.5.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.2.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.1.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.3.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.2.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.2.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.5.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.3.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.3.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.3.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.3.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.2.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.5.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.3.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.self_attn.q_proj.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.4.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.3.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.3.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.self_attn.k_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.self_attn.q_proj.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.self_attn.o_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.self_attn.q_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.self_attn.v_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.self_attn.k_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.2.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.6.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.6.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.4.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.3.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.4.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.4.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.4.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.4.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.6.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.4.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.0.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.1.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.3.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.5.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.5.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.0.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.3.self_attn.v_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.4.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.2.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.5.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.5.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.9.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.9.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.9.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.9.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.8.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.8.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.8.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.9.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.11.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.10.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.13.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.12.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.9.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.7.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.12.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.self_attn.o_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.12.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.13.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.12.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.self_attn.o_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.9.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.self_attn.k_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.self_attn.v_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.self_attn.q_proj.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.10.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.13.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.11.self_attn.k_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.10.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.self_attn.q_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.self_attn.k_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.8.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.self_attn.o_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.12.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.11.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.10.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.10.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.11.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.self_attn.v_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.11.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.11.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.10.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.11.self_attn.v_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.10.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.8.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.7.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.10.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.12.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.9.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.12.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.11.self_attn.q_proj.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.10.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.12.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.12.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.10.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.10.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.8.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.self_attn.k_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.9.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.11.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.11.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.12.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.11.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.11.self_attn.q_proj.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.10.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.10.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.7.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.11.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.13.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.11.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.12.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.8.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.14.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.7.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.11.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.13.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.10.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.13.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.13.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.8.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.13.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.7.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.13.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.14.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.11.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.14.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.14.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.14.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.14.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.13.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.8.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.10.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.13.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.7.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.7.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.7.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.8.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.8.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.7.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.12.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.8.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.14.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.8.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.9.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.8.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.10.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.14.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.12.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.8.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.8.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.22.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.22.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.22.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.22.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.18.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.14.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.19.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.17.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.16.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.16.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.17.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.19.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.17.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.14.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.17.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.19.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.self_attn.v_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.16.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.16.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.16.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.15.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.22.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.15.self_attn.k_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.15.self_attn.k_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.14.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.15.self_attn.o_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.15.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.15.self_attn.k_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.self_attn.q_proj.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.15.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.17.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.16.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.16.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.17.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.16.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.17.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.16.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.17.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.17.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.self_attn.k_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.19.self_attn.k_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.14.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.18.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.18.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.15.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.self_attn.k_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.16.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.self_attn.o_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.self_attn.q_proj.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.self_attn.v_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.self_attn.q_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.self_attn.q_proj.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.16.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.22.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.17.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.17.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.19.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.self_attn.q_proj.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.17.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.21.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.18.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.18.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.19.self_attn.q_proj.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.18.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.self_attn.o_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.18.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.18.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.21.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.18.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.self_attn.o_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.18.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.18.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.19.self_attn.q_proj.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.self_attn.v_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.18.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.16.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.19.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.17.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.19.self_attn.k_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.self_attn.k_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.16.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.16.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.self_attn.o_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.self_attn.o_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.19.self_attn.q_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.15.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.self_attn.v_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.self_attn.v_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.19.self_attn.v_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.17.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.20.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.15.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.19.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.20.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.29.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.29.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.29.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.29.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.29.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.29.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.29.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.29.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.28.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.28.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.28.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.28.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.28.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.25.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.28.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.28.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.self_attn.k_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.28.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.25.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.26.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.28.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.24.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.25.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.self_attn.v_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.24.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.27.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.24.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.24.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.25.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.24.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.22.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.25.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.24.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.24.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.self_attn.v_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.22.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.24.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.24.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.29.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.25.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.27.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.self_attn.o_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.self_attn.k_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.self_attn.k_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.24.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.23.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.self_attn.q_proj.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.24.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.23.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.self_attn.o_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.24.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.22.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.25.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.self_attn.v_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.28.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.self_attn.q_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.22.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.23.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.self_attn.q_proj.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.29.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.23.self_attn.o_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.25.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.22.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.self_attn.v_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.24.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.23.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.24.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.23.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.29.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.self_attn.k_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.23.self_attn.k_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.self_attn.o_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.25.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.23.self_attn.q_proj.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.self_attn.v_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.25.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.25.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.25.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.27.self_attn.q_proj.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.25.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.23.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.26.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.self_attn.v_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.27.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.26.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.26.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.26.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.26.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.26.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.26.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.self_attn.o_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.26.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.26.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.26.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.26.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.29.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.23.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.27.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.27.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.27.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.24.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.self_attn.k_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.27.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.27.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.27.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.self_attn.k_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.27.self_attn.k_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.29.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.self_attn.o_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.27.self_attn.q_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.self_attn.q_proj.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.27.self_attn.q_proj.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.27.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.26.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.29.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.28.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.22.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.25.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.28.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.28.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.25.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.28.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.37.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.37.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.37.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.37.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.37.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.37.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.37.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.36.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.36.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.36.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.36.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.36.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.36.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.32.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.33.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.36.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.33.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.32.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.32.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.32.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.33.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.32.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.34.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.32.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.37.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.31.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.self_attn.k_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.self_attn.k_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.self_attn.k_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.32.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.self_attn.v_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.33.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.self_attn.q_proj.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.31.self_attn.k_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.36.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.self_attn.v_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.35.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.self_attn.v_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.35.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.32.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.32.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.32.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.33.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.35.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.31.self_attn.o_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.31.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.36.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.33.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.32.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.32.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.33.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.33.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.36.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.31.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.37.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.self_attn.o_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.32.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.31.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.self_attn.v_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.34.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.32.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.31.self_attn.q_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.self_attn.k_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.31.self_attn.q_proj.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.33.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.self_attn.q_proj.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.35.self_attn.o_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.33.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.33.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.34.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.35.self_attn.q_proj.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.35.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.34.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.30.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.34.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.32.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.34.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.34.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.35.self_attn.o_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.36.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.34.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.33.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.37.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.self_attn.o_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.35.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.35.self_attn.k_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.31.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.35.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.30.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.36.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.self_attn.k_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.self_attn.k_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.35.self_attn.o_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.self_attn.q_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.33.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.36.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.35.self_attn.q_proj.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.self_attn.q_proj.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.31.self_attn.v_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.35.self_attn.v_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.34.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.36.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.33.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.34.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.35.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.36.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.self_attn.v_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.self_attn.q_proj.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.self_attn.q_proj.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.self_attn.q_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.self_attn.o_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.self_attn.o_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.self_attn.k_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.39.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.norm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.39.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.39.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.self_attn.k_norm.weight",
      "shape": [
        256
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.39.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.39.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.39.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.self_attn.v_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.post_attention_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.switch_mlp.up_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.38.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.linear_attn.in_proj_b.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.38.linear_attn.in_proj_b.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.38.linear_attn.in_proj_qkv.biases",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.linear_attn.in_proj_a.scales",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.linear_attn.in_proj_a.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.linear_attn.A_log",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.linear_attn.conv1d.weight",
      "shape": [
        8192,
        4,
        1
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.38.linear_attn.in_proj_b.biases",
      "shape": [
        32,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.lm_head.biases",
      "shape": [
        248320,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.lm_head.weight",
      "shape": [
        248320,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.38.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.37.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.37.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.37.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.linear_attn.out_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.38.linear_attn.dt_bias",
      "shape": [
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.linear_attn.in_proj_z.scales",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.37.mlp.shared_expert.up_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.self_attn.q_proj.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.38.linear_attn.in_proj_a.weight",
      "shape": [
        32,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.39.self_attn.o_proj.weight",
      "shape": [
        2048,
        768
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.37.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.input_layernorm.weight",
      "shape": [
        2048
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.mlp.switch_mlp.up_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.self_attn.v_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.38.linear_attn.in_proj_z.biases",
      "shape": [
        4096,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.37.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.gate.biases",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.self_attn.k_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.mlp.shared_expert.up_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.38.mlp.switch_mlp.gate_proj.weight",
      "shape": [
        256,
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.38.linear_attn.in_proj_qkv.scales",
      "shape": [
        8192,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.switch_mlp.up_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.linear_attn.in_proj_qkv.weight",
      "shape": [
        8192,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.38.linear_attn.in_proj_z.weight",
      "shape": [
        4096,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.39.mlp.shared_expert.down_proj.biases",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.linear_attn.norm.weight",
      "shape": [
        128
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.linear_attn.out_proj.scales",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.mlp.shared_expert_gate.scales",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.gate.scales",
      "shape": [
        256,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.39.self_attn.k_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.38.mlp.gate.weight",
      "shape": [
        256,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.39.mlp.shared_expert.gate_proj.weight",
      "shape": [
        512,
        384
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.38.linear_attn.out_proj.biases",
      "shape": [
        2048,
        64
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.shared_expert_gate.weight",
      "shape": [
        1,
        512
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.lm_head.scales",
      "shape": [
        248320,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.shared_expert.down_proj.scales",
      "shape": [
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.shared_expert.down_proj.weight",
      "shape": [
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.39.mlp.switch_mlp.gate_proj.biases",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.shared_expert.gate_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.shared_expert.gate_proj.scales",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.shared_expert.up_proj.biases",
      "shape": [
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.shared_expert_gate.biases",
      "shape": [
        1,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.37.mlp.switch_mlp.down_proj.scales",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.switch_mlp.down_proj.biases",
      "shape": [
        256,
        2048,
        8
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    },
    {
      "name": "language_model.model.layers.38.mlp.switch_mlp.down_proj.weight",
      "shape": [
        256,
        2048,
        96
      ],
      "dtype": "mlx.core.uint32"
    },
    {
      "name": "language_model.model.layers.38.mlp.switch_mlp.gate_proj.scales",
      "shape": [
        256,
        512,
        32
      ],
      "dtype": "mlx.core.bfloat16",
      "all_stored_values_finite": true
    }
  ],
  "generation": [
    {
      "id": "text_en",
      "messages": [
        {
          "role": "user",
          "content": "Explain in two short sentences why dividing a job into smaller tasks can help a team."
        }
      ],
      "content": "Dividing a job into smaller tasks allows team members to specialize and work on specific components simultaneously, which significantly boosts overall efficiency. This approach also clarifies individual responsibilities, making it easier to track progress and manage deadlines.",
      "expected": null,
      "passed": true,
      "finish_reason": "stop",
      "tokens": 44,
      "full_vocab_finite_logit_checks": 45,
      "seconds": 99.09510660171509
    },
    {
      "id": "text_zh",
      "messages": [
        {
          "role": "user",
          "content": "请用两句话说明为什么重要文件需要备份。"
        }
      ],
      "content": "备份能确保在硬件故障、人为误删或恶意攻击导致数据丢失时，关键信息依然完好无损。它是应对不可预见灾难的最后防线，保障业务连续性和个人记忆的安全。",
      "expected": null,
      "passed": true,
      "finish_reason": "stop",
      "tokens": 41,
      "full_vocab_finite_logit_checks": 42,
      "seconds": 1.326098918914795
    },
    {
      "id": "math_1",
      "messages": [
        {
          "role": "user",
          "content": "Return only the integer: 37 + 58."
        }
      ],
      "content": "95",
      "expected": "95",
      "passed": true,
      "finish_reason": "stop",
      "tokens": 3,
      "full_vocab_finite_logit_checks": 4,
      "seconds": 0.7312948703765869
    },
    {
      "id": "math_2",
      "messages": [
        {
          "role": "user",
          "content": "Return only the integer: 144 divided by 12."
        }
      ],
      "content": "12",
      "expected": "12",
      "passed": true,
      "finish_reason": "stop",
      "tokens": 3,
      "full_vocab_finite_logit_checks": 4,
      "seconds": 0.5915679931640625
    },
    {
      "id": "json_1",
      "messages": [
        {
          "role": "user",
          "content": "Return only valid JSON with no markdown. The object must be exactly {\"city\":\"Paris\",\"count\":3}."
        }
      ],
      "content": "{\"city\":\"Paris\",\"count\":3}",
      "expected": {
        "city": "Paris",
        "count": 3
      },
      "passed": true,
      "finish_reason": "stop",
      "tokens": 10,
      "full_vocab_finite_logit_checks": 11,
      "seconds": 54.92182898521423
    },
    {
      "id": "json_2",
      "messages": [
        {
          "role": "user",
          "content": "Return only valid JSON with no markdown: an array of the even integers from 2 through 8, in increasing order."
        }
      ],
      "content": "[2, 4, 6, 8]",
      "expected": [
        2,
        4,
        6,
        8
      ],
      "passed": true,
      "finish_reason": "stop",
      "tokens": 13,
      "full_vocab_finite_logit_checks": 14,
      "seconds": 1.7010161876678467
    },
    {
      "id": "memory_en",
      "messages": [
        {
          "role": "user",
          "content": "Remember the code word: CORAL-27."
        },
        {
          "role": "assistant",
          "content": "The code word is CORAL-27."
        },
        {
          "role": "user",
          "content": "Return only the code word I gave you."
        }
      ],
      "content": "CORAL-27",
      "expected": "CORAL-27",
      "passed": true,
      "finish_reason": "stop",
      "tokens": 6,
      "full_vocab_finite_logit_checks": 7,
      "seconds": 0.939906120300293
    },
    {
      "id": "memory_zh",
      "messages": [
        {
          "role": "user",
          "content": "记住：仓库编号是海风42。"
        },
        {
          "role": "assistant",
          "content": "仓库编号是海风42。"
        },
        {
          "role": "user",
          "content": "只回答刚才的仓库编号。"
        }
      ],
      "content": "海风42",
      "expected": "海风42",
      "passed": true,
      "finish_reason": "stop",
      "tokens": 5,
      "full_vocab_finite_logit_checks": 6,
      "seconds": 0.8742678165435791
    }
  ],
  "all_stored_floating_values_finite": true,
  "weight_bytes": 28168623119,
  "strict_stock_reload": true,
  "quantized_modules": [
    {
      "module": "language_model.lm_head",
      "bits": 6,
      "rows_checked": 248320,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.embed_tokens",
      "bits": 6,
      "rows_checked": 248320,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.0.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.0.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.0.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.0.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.0.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.0.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.0.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.0.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.0.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.0.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.0.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.0.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.0.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.1.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.1.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.1.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.1.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.1.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.1.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.1.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.1.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.1.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.1.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.1.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.1.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.1.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.10.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.10.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.10.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.10.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.10.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.10.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.10.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.10.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.10.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.10.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.10.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.10.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.10.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.11.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.11.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.11.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.11.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.11.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.11.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.11.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.11.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.11.self_attn.k_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.11.self_attn.o_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.11.self_attn.q_proj",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.11.self_attn.v_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.12.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.12.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.12.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.12.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.12.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.12.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.12.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.12.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.12.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.12.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.12.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.12.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.12.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.13.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.13.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.13.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.13.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.13.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.13.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.13.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.13.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.13.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.13.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.13.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.13.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.13.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.14.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.14.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.14.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.14.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.14.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.14.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.14.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.14.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.14.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.14.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.14.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.14.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.14.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.15.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.15.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.15.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.15.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.15.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.15.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.15.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.15.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.15.self_attn.k_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.15.self_attn.o_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.15.self_attn.q_proj",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.15.self_attn.v_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.16.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.16.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.16.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.16.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.16.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.16.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.16.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.16.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.16.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.16.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.16.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.16.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.16.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.17.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.17.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.17.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.17.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.17.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.17.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.17.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.17.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.17.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.17.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.17.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.17.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.17.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.18.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.18.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.18.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.18.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.18.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.18.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.18.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.18.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.18.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.18.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.18.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.18.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.18.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.19.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.19.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.19.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.19.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.19.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.19.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.19.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.19.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.19.self_attn.k_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.19.self_attn.o_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.19.self_attn.q_proj",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.19.self_attn.v_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.2.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.2.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.2.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.2.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.2.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.2.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.2.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.2.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.2.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.2.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.2.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.2.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.2.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.20.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.20.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.20.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.20.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.20.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.20.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.20.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.20.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.20.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.20.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.20.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.20.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.20.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.21.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.21.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.21.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.21.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.21.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.21.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.21.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.21.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.21.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.21.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.21.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.21.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.21.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.22.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.22.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.22.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.22.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.22.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.22.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.22.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.22.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.22.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.22.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.22.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.22.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.22.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.23.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.23.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.23.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.23.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.23.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.23.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.23.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.23.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.23.self_attn.k_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.23.self_attn.o_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.23.self_attn.q_proj",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.23.self_attn.v_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.24.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.24.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.24.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.24.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.24.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.24.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.24.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.24.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.24.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.24.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.24.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.24.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.24.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.25.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.25.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.25.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.25.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.25.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.25.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.25.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.25.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.25.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.25.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.25.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.25.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.25.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.26.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.26.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.26.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.26.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.26.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.26.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.26.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.26.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.26.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.26.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.26.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.26.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.26.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.27.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.27.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.27.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.27.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.27.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.27.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.27.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.27.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.27.self_attn.k_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.27.self_attn.o_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.27.self_attn.q_proj",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.27.self_attn.v_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.28.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.28.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.28.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.28.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.28.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.28.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.28.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.28.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.28.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.28.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.28.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.28.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.28.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.29.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.29.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.29.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.29.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.29.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.29.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.29.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.29.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.29.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.29.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.29.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.29.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.29.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.3.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.3.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.3.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.3.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.3.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.3.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.3.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.3.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.3.self_attn.k_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.3.self_attn.o_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.3.self_attn.q_proj",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.3.self_attn.v_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.30.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.30.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.30.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.30.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.30.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.30.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.30.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.30.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.30.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.30.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.30.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.30.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.30.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.31.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.31.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.31.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.31.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.31.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.31.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.31.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.31.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.31.self_attn.k_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.31.self_attn.o_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.31.self_attn.q_proj",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.31.self_attn.v_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.32.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.32.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.32.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.32.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.32.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.32.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.32.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.32.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.32.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.32.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.32.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.32.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.32.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.33.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.33.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.33.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.33.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.33.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.33.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.33.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.33.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.33.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.33.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.33.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.33.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.33.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.34.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.34.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.34.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.34.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.34.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.34.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.34.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.34.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.34.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.34.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.34.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.34.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.34.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.35.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.35.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.35.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.35.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.35.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.35.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.35.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.35.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.35.self_attn.k_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.35.self_attn.o_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.35.self_attn.q_proj",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.35.self_attn.v_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.36.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.36.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.36.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.36.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.36.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.36.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.36.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.36.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.36.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.36.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.36.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.36.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.36.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.37.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.37.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.37.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.37.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.37.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.37.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.37.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.37.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.37.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.37.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.37.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.37.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.37.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.38.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.38.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.38.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.38.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.38.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.38.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.38.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.38.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.38.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.38.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.38.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.38.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.38.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.39.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.39.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.39.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.39.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.39.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.39.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.39.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.39.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.39.self_attn.k_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.39.self_attn.o_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.39.self_attn.q_proj",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.39.self_attn.v_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.4.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.4.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.4.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.4.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.4.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.4.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.4.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.4.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.4.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.4.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.4.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.4.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.4.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.5.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.5.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.5.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.5.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.5.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.5.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.5.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.5.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.5.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.5.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.5.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.5.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.5.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.6.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.6.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.6.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.6.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.6.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.6.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.6.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.6.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.6.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.6.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.6.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.6.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.6.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.7.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.7.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.7.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.7.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.7.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.7.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.7.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.7.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.7.self_attn.k_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.7.self_attn.o_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.7.self_attn.q_proj",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.7.self_attn.v_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.8.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.8.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.8.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.8.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.8.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.8.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.8.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.8.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.8.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.8.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.8.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.8.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.8.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.9.linear_attn.in_proj_a",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.9.linear_attn.in_proj_b",
      "bits": 6,
      "rows_checked": 32,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.9.linear_attn.in_proj_qkv",
      "bits": 6,
      "rows_checked": 8192,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.9.linear_attn.in_proj_z",
      "bits": 6,
      "rows_checked": 4096,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.9.linear_attn.out_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 4096,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.9.mlp.gate",
      "bits": 8,
      "rows_checked": 256,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.9.mlp.shared_expert.down_proj",
      "bits": 6,
      "rows_checked": 2048,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.9.mlp.shared_expert.gate_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.9.mlp.shared_expert.up_proj",
      "bits": 6,
      "rows_checked": 512,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.9.mlp.shared_expert_gate",
      "bits": 8,
      "rows_checked": 1,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.9.mlp.switch_mlp.down_proj",
      "bits": 6,
      "rows_checked": 524288,
      "columns": 512,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.9.mlp.switch_mlp.gate_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    },
    {
      "module": "language_model.model.layers.9.mlp.switch_mlp.up_proj",
      "bits": 6,
      "rows_checked": 131072,
      "columns": 2048,
      "all_dequantized_values_finite": true
    }
  ],
  "all_dequantized_values_finite": true,
  "tokenizer_source_files": {
    "tokenizer.json": "06b9509352d2af50381ab2247e083b80d32d5c0aba91c272ca9ff729b6a0e523",
    "tokenizer_config.json": "792fa3f0cb88b111e54ef3134c873531008c4df471d108da17903426e308aa7b",
    "chat_template.jinja": "e84f32a23fdda27689f868aa4a1a5621f41133e51a48d7f3efcbea2839574259"
  },
  "functional_passed": 8,
  "functional_total": 8,
  "all_generation_logits_finite": true
}