config.json

{
  "_name_or_path": "RuadaptQwen2.5-14B-Instruct-1M-AWQ-64G-INT4-vGEMM",
  "architectures": [
    "Qwen2ForCausalLM"
  ],
  "attention_dropout": 0.0,
  "dual_chunk_attention_config": {
    "chunk_size": 262144,
    "local_size": 8192,
    "original_max_position_embeddings": 262144
  },
  "eos_token_id": 145111,
  "hidden_act": "silu",
  "hidden_size": 5120,
  "initializer_range": 0.02,
  "intermediate_size": 13824,
  "max_position_embeddings": 1010000,
  "max_window_layers": 48,
  "model_type": "qwen2",
  "num_attention_heads": 40,
  "num_hidden_layers": 48,
  "num_key_value_heads": 8,
  "pad_token_id": 145109,
  "quantization_config": {
    "backend": "autoawq",
    "bits": 4,
    "do_fuse": false,
    "exllama_config": null,
    "fuse_max_seq_len": null,
    "group_size": 64,
    "modules_to_fuse": null,
    "modules_to_not_convert": null,
    "quant_method": "awq",
    "version": "gemm",
    "zero_point": true
  },
  "rms_norm_eps": 1e-05,
  "rope_scaling": null,
  "rope_theta": 10000000.0,
  "sliding_window": null,
  "tie_word_embeddings": false,
  "torch_dtype": "float16",
  "transformers_version": "4.47.1",
  "use_cache": false,
  "use_sliding_window": false,
  "vocab_size": 145152
}