config.json

{
  "architectures": [
    "Qwen3_5ForCausalLM"
  ],
  "model_type": "qwen35",
  "vocab_size": 248320,
  "hidden_size": 5120,
  "intermediate_size": 17408,
  "num_hidden_layers": 65,
  "num_attention_heads": 24,
  "num_key_value_heads": 4,
  "head_dim": 256,
  "max_position_embeddings": 262144,
  "rope_theta": 10000000,
  "rms_norm_eps": 1e-6,
  "hidden_act": "silu",
  "tie_word_embeddings": false,
  "use_cache": true,

  "linear_attn_config": {
    "conv_kernel_size": 4,
    "state_size": 128,
    "num_groups": 16,
    "time_step_rank": 48,
    "inner_size": 6144
  },

  "bos_token_id": 248044,
  "pad_token_id": 248044,
  "eos_token_id": [
    248046,
    248044
  ],

  "torch_dtype": "bfloat16",
  "transformers_version": "4.51.0"
}