Inkling-2-Small-NVFP4-CSF / hf_quant_config.json
original-el8's picture
Add files using upload-large-folder tool
af3c078 verified
Raw History Blame Contribute Delete
15.8 kB
{
"producer": {
"name": "modelopt",
"version": "0.48.0rc1.dev355+gcc6963dea"
},
"quantization": {
"quant_algo": "NVFP4",
"kv_cache_quant_algo": null,
"group_size": 16,
"exclude_modules": [
"lm_head",
"model.vision_tower*",
"model.language_model.layers.0.mlp.gate",
"model.language_model.layers.0.self_attn*",
"model.language_model.layers.0.post_attention_convolution*",
"model.language_model.layers.0.post_mlp_convolution*",
"model.language_model.layers.1.mlp.gate",
"model.language_model.layers.1.self_attn*",
"model.language_model.layers.1.post_attention_convolution*",
"model.language_model.layers.1.post_mlp_convolution*",
"model.language_model.layers.2.mlp.gate",
"model.language_model.layers.2.self_attn*",
"model.language_model.layers.2.post_attention_convolution*",
"model.language_model.layers.2.post_mlp_convolution*",
"model.language_model.layers.3.mlp.gate",
"model.language_model.layers.3.self_attn*",
"model.language_model.layers.3.post_attention_convolution*",
"model.language_model.layers.3.post_mlp_convolution*",
"model.language_model.layers.4.mlp.gate",
"model.language_model.layers.4.self_attn*",
"model.language_model.layers.4.post_attention_convolution*",
"model.language_model.layers.4.post_mlp_convolution*",
"model.language_model.layers.5.mlp.gate",
"model.language_model.layers.5.self_attn*",
"model.language_model.layers.5.post_attention_convolution*",
"model.language_model.layers.5.post_mlp_convolution*",
"model.language_model.layers.6.mlp.gate",
"model.language_model.layers.6.self_attn*",
"model.language_model.layers.6.post_attention_convolution*",
"model.language_model.layers.6.post_mlp_convolution*",
"model.language_model.layers.7.mlp.gate",
"model.language_model.layers.7.self_attn*",
"model.language_model.layers.7.post_attention_convolution*",
"model.language_model.layers.7.post_mlp_convolution*",
"model.language_model.layers.8.mlp.gate",
"model.language_model.layers.8.self_attn*",
"model.language_model.layers.8.post_attention_convolution*",
"model.language_model.layers.8.post_mlp_convolution*",
"model.language_model.layers.9.mlp.gate",
"model.language_model.layers.9.self_attn*",
"model.language_model.layers.9.post_attention_convolution*",
"model.language_model.layers.9.post_mlp_convolution*",
"model.language_model.layers.10.mlp.gate",
"model.language_model.layers.10.self_attn*",
"model.language_model.layers.10.post_attention_convolution*",
"model.language_model.layers.10.post_mlp_convolution*",
"model.language_model.layers.11.mlp.gate",
"model.language_model.layers.11.self_attn*",
"model.language_model.layers.11.post_attention_convolution*",
"model.language_model.layers.11.post_mlp_convolution*",
"model.language_model.layers.12.mlp.gate",
"model.language_model.layers.12.self_attn*",
"model.language_model.layers.12.post_attention_convolution*",
"model.language_model.layers.12.post_mlp_convolution*",
"model.language_model.layers.13.mlp.gate",
"model.language_model.layers.13.self_attn*",
"model.language_model.layers.13.post_attention_convolution*",
"model.language_model.layers.13.post_mlp_convolution*",
"model.language_model.layers.14.mlp.gate",
"model.language_model.layers.14.self_attn*",
"model.language_model.layers.14.post_attention_convolution*",
"model.language_model.layers.14.post_mlp_convolution*",
"model.language_model.layers.15.mlp.gate",
"model.language_model.layers.15.self_attn*",
"model.language_model.layers.15.post_attention_convolution*",
"model.language_model.layers.15.post_mlp_convolution*",
"model.language_model.layers.16.mlp.gate",
"model.language_model.layers.16.self_attn*",
"model.language_model.layers.16.post_attention_convolution*",
"model.language_model.layers.16.post_mlp_convolution*",
"model.language_model.layers.17.mlp.gate",
"model.language_model.layers.17.self_attn*",
"model.language_model.layers.17.post_attention_convolution*",
"model.language_model.layers.17.post_mlp_convolution*",
"model.language_model.layers.18.mlp.gate",
"model.language_model.layers.18.self_attn*",
"model.language_model.layers.18.post_attention_convolution*",
"model.language_model.layers.18.post_mlp_convolution*",
"model.language_model.layers.19.mlp.gate",
"model.language_model.layers.19.self_attn*",
"model.language_model.layers.19.post_attention_convolution*",
"model.language_model.layers.19.post_mlp_convolution*",
"model.language_model.layers.20.mlp.gate",
"model.language_model.layers.20.self_attn*",
"model.language_model.layers.20.post_attention_convolution*",
"model.language_model.layers.20.post_mlp_convolution*",
"model.language_model.layers.21.mlp.gate",
"model.language_model.layers.21.self_attn*",
"model.language_model.layers.21.post_attention_convolution*",
"model.language_model.layers.21.post_mlp_convolution*",
"model.language_model.layers.22.mlp.gate",
"model.language_model.layers.22.self_attn*",
"model.language_model.layers.22.post_attention_convolution*",
"model.language_model.layers.22.post_mlp_convolution*",
"model.language_model.layers.23.mlp.gate",
"model.language_model.layers.23.self_attn*",
"model.language_model.layers.23.post_attention_convolution*",
"model.language_model.layers.23.post_mlp_convolution*",
"model.language_model.layers.24.mlp.gate",
"model.language_model.layers.24.self_attn*",
"model.language_model.layers.24.post_attention_convolution*",
"model.language_model.layers.24.post_mlp_convolution*",
"model.language_model.layers.25.mlp.gate",
"model.language_model.layers.25.self_attn*",
"model.language_model.layers.25.post_attention_convolution*",
"model.language_model.layers.25.post_mlp_convolution*",
"model.language_model.layers.26.mlp.gate",
"model.language_model.layers.26.self_attn*",
"model.language_model.layers.26.post_attention_convolution*",
"model.language_model.layers.26.post_mlp_convolution*",
"model.language_model.layers.27.mlp.gate",
"model.language_model.layers.27.self_attn*",
"model.language_model.layers.27.post_attention_convolution*",
"model.language_model.layers.27.post_mlp_convolution*",
"model.language_model.layers.28.mlp.gate",
"model.language_model.layers.28.self_attn*",
"model.language_model.layers.28.post_attention_convolution*",
"model.language_model.layers.28.post_mlp_convolution*",
"model.language_model.layers.29.mlp.gate",
"model.language_model.layers.29.self_attn*",
"model.language_model.layers.29.post_attention_convolution*",
"model.language_model.layers.29.post_mlp_convolution*",
"model.language_model.layers.30.mlp.gate",
"model.language_model.layers.30.self_attn*",
"model.language_model.layers.30.post_attention_convolution*",
"model.language_model.layers.30.post_mlp_convolution*",
"model.language_model.layers.31.mlp.gate",
"model.language_model.layers.31.self_attn*",
"model.language_model.layers.31.post_attention_convolution*",
"model.language_model.layers.31.post_mlp_convolution*",
"model.language_model.layers.32.mlp.gate",
"model.language_model.layers.32.self_attn*",
"model.language_model.layers.32.post_attention_convolution*",
"model.language_model.layers.32.post_mlp_convolution*",
"model.language_model.layers.33.mlp.gate",
"model.language_model.layers.33.self_attn*",
"model.language_model.layers.33.post_attention_convolution*",
"model.language_model.layers.33.post_mlp_convolution*",
"model.language_model.layers.34.mlp.gate",
"model.language_model.layers.34.self_attn*",
"model.language_model.layers.34.post_attention_convolution*",
"model.language_model.layers.34.post_mlp_convolution*",
"model.language_model.layers.35.mlp.gate",
"model.language_model.layers.35.self_attn*",
"model.language_model.layers.35.post_attention_convolution*",
"model.language_model.layers.35.post_mlp_convolution*",
"model.language_model.layers.36.mlp.gate",
"model.language_model.layers.36.self_attn*",
"model.language_model.layers.36.post_attention_convolution*",
"model.language_model.layers.36.post_mlp_convolution*",
"model.language_model.layers.37.mlp.gate",
"model.language_model.layers.37.self_attn*",
"model.language_model.layers.37.post_attention_convolution*",
"model.language_model.layers.37.post_mlp_convolution*",
"model.language_model.layers.38.mlp.gate",
"model.language_model.layers.38.self_attn*",
"model.language_model.layers.38.post_attention_convolution*",
"model.language_model.layers.38.post_mlp_convolution*",
"model.language_model.layers.39.mlp.gate",
"model.language_model.layers.39.self_attn*",
"model.language_model.layers.39.post_attention_convolution*",
"model.language_model.layers.39.post_mlp_convolution*",
"model.language_model.layers.40.mlp.gate",
"model.language_model.layers.40.self_attn*",
"model.language_model.layers.40.post_attention_convolution*",
"model.language_model.layers.40.post_mlp_convolution*",
"model.language_model.layers.41.mlp.gate",
"model.language_model.layers.41.self_attn*",
"model.language_model.layers.41.post_attention_convolution*",
"model.language_model.layers.41.post_mlp_convolution*",
"model.language_model.layers.42.mlp.gate",
"model.language_model.layers.42.self_attn*",
"model.language_model.layers.42.post_attention_convolution*",
"model.language_model.layers.42.post_mlp_convolution*",
"model.language_model.layers.43.mlp.gate",
"model.language_model.layers.43.self_attn*",
"model.language_model.layers.43.post_attention_convolution*",
"model.language_model.layers.43.post_mlp_convolution*",
"model.language_model.layers.44.mlp.gate",
"model.language_model.layers.44.self_attn*",
"model.language_model.layers.44.post_attention_convolution*",
"model.language_model.layers.44.post_mlp_convolution*",
"model.language_model.layers.45.mlp.gate",
"model.language_model.layers.45.self_attn*",
"model.language_model.layers.45.post_attention_convolution*",
"model.language_model.layers.45.post_mlp_convolution*",
"model.language_model.layers.46.mlp.gate",
"model.language_model.layers.46.self_attn*",
"model.language_model.layers.46.post_attention_convolution*",
"model.language_model.layers.46.post_mlp_convolution*",
"model.language_model.layers.47.mlp.gate",
"model.language_model.layers.47.self_attn*",
"model.language_model.layers.47.post_attention_convolution*",
"model.language_model.layers.47.post_mlp_convolution*"
],
"weight_scale_encoding": "csf_32x2_v1"
},
"mtp": {
"parameter_groups": [
{
"compute_dtype": "bfloat16",
"output_dtype": "bfloat16",
"required_activation_quantization": null,
"tensors": [
"fuse_weight",
"layers.0.dflare_norm.weight",
"layers.0.input_layernorm.weight",
"layers.0.post_attention_layernorm.weight",
"layers.0.self_attn.k_norm.weight",
"layers.0.self_attn.k_norm_ctx.weight",
"layers.0.self_attn.k_proj.weight",
"layers.0.self_attn.k_proj_ctx.weight",
"layers.0.self_attn.o_proj.weight",
"layers.0.self_attn.q_norm.weight",
"layers.0.self_attn.q_proj.weight",
"layers.0.self_attn.v_proj.weight",
"layers.0.self_attn.v_proj_ctx.weight",
"layers.1.dflare_norm.weight",
"layers.1.input_layernorm.weight",
"layers.1.post_attention_layernorm.weight",
"layers.1.self_attn.k_norm.weight",
"layers.1.self_attn.k_norm_ctx.weight",
"layers.1.self_attn.k_proj.weight",
"layers.1.self_attn.k_proj_ctx.weight",
"layers.1.self_attn.o_proj.weight",
"layers.1.self_attn.q_norm.weight",
"layers.1.self_attn.q_proj.weight",
"layers.1.self_attn.v_proj.weight",
"layers.1.self_attn.v_proj_ctx.weight",
"layers.2.dflare_norm.weight",
"layers.2.input_layernorm.weight",
"layers.2.post_attention_layernorm.weight",
"layers.2.self_attn.k_norm.weight",
"layers.2.self_attn.k_norm_ctx.weight",
"layers.2.self_attn.k_proj.weight",
"layers.2.self_attn.k_proj_ctx.weight",
"layers.2.self_attn.o_proj.weight",
"layers.2.self_attn.q_norm.weight",
"layers.2.self_attn.q_proj.weight",
"layers.2.self_attn.v_proj.weight",
"layers.2.self_attn.v_proj_ctx.weight",
"layers.3.dflare_norm.weight",
"layers.3.input_layernorm.weight",
"layers.3.post_attention_layernorm.weight",
"layers.3.self_attn.k_norm.weight",
"layers.3.self_attn.k_norm_ctx.weight",
"layers.3.self_attn.k_proj.weight",
"layers.3.self_attn.k_proj_ctx.weight",
"layers.3.self_attn.o_proj.weight",
"layers.3.self_attn.q_norm.weight",
"layers.3.self_attn.q_proj.weight",
"layers.3.self_attn.v_proj.weight",
"layers.3.self_attn.v_proj_ctx.weight",
"layers.4.dflare_norm.weight",
"layers.4.input_layernorm.weight",
"layers.4.post_attention_layernorm.weight",
"layers.4.self_attn.k_norm.weight",
"layers.4.self_attn.k_norm_ctx.weight",
"layers.4.self_attn.k_proj.weight",
"layers.4.self_attn.k_proj_ctx.weight",
"layers.4.self_attn.o_proj.weight",
"layers.4.self_attn.q_norm.weight",
"layers.4.self_attn.q_proj.weight",
"layers.4.self_attn.v_proj.weight",
"layers.4.self_attn.v_proj_ctx.weight",
"markov_head.dflare_norm.weight",
"markov_head.hidden_out.bias",
"markov_head.hidden_out.weight",
"markov_head.markov_w1.weight",
"markov_head.rnn_proj.bias",
"markov_head.rnn_proj.weight",
"mask_embed",
"norm.weight"
],
"weight_block_shape": [],
"weight_transform": "cast"
},
{
"compute_dtype": "mxfp8",
"output_dtype": "float8_e4m3fn",
"required_activation_quantization": {
"block_size": 32,
"format": "mxfp8_e4m3",
"rounding": "nearest",
"scale_dtype": "ue8m0"
},
"tensors": [
"layers.0.mlp.down_proj.weight",
"layers.0.mlp.gate_proj.weight",
"layers.0.mlp.up_proj.weight",
"layers.1.mlp.down_proj.weight",
"layers.1.mlp.gate_proj.weight",
"layers.1.mlp.up_proj.weight",
"layers.2.mlp.down_proj.weight",
"layers.2.mlp.gate_proj.weight",
"layers.2.mlp.up_proj.weight",
"layers.3.mlp.down_proj.weight",
"layers.3.mlp.gate_proj.weight",
"layers.3.mlp.up_proj.weight",
"layers.4.mlp.down_proj.weight",
"layers.4.mlp.gate_proj.weight",
"layers.4.mlp.up_proj.weight"
],
"weight_block_shape": [
1,
32
],
"weight_transform": "mxfp8_nearest"
}
]
}
}