config.json

{
  "_name_or_path": "/slot/sandbox/j/_tmp/datanre1owck/hf_model",
  "architectures": [
    "Qwen2ForCausalLM"
  ],
  "attention_dropout": 0.0,
  "bos_token_id": 151643,
  "eos_token_id": 151645,
  "hidden_act": "silu",
  "hidden_size": 8192,
  "initializer_range": 0.02,
  "intermediate_size": 29568,
  "max_position_embeddings": 32768,
  "max_window_layers": 80,
  "model_type": "qwen2",
  "num_attention_heads": 64,
  "num_hidden_layers": 80,
  "num_key_value_heads": 8,
  "quantization_config": {
    "in_group_size": 8,
    "linear_weights_not_to_quantize": [
      "lm_head.weight"
    ],
    "nbits_per_codebook": 16,
    "num_codebooks": 1,
    "out_group_size": 1,
    "quant_method": "aqlm"
  },
  "rms_norm_eps": 1e-06,
  "rope_theta": 1000000.0,
  "sliding_window": 131072,
  "tie_word_embeddings": false,
  "torch_dtype": "float16",
  "transformers_version": "4.40.1",
  "use_cache": true,
  "use_sliding_window": false,
  "vocab_size": 152064
}