|
{
|
|
"_name_or_path": "C:/Qwen2.5-Coder-14B-YOYO-SCE-1M",
|
|
"architectures": [
|
|
"Qwen2ForCausalLM"
|
|
],
|
|
"attention_dropout": 0.0,
|
|
"bos_token_id": 151643,
|
|
"dual_chunk_attention_config": {
|
|
"chunk_size": 262144,
|
|
"local_size": 8192,
|
|
"original_max_position_embeddings": 262144
|
|
},
|
|
"eos_token_id": 151645,
|
|
"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,
|
|
"rms_norm_eps": 1e-05,
|
|
"rope_scaling": null,
|
|
"rope_theta": 10000000.0,
|
|
"sliding_window": null,
|
|
"tie_word_embeddings": false,
|
|
"torch_dtype": "bfloat16",
|
|
"transformers_version": "4.48.1",
|
|
"use_cache": true,
|
|
"use_sliding_window": false,
|
|
"vocab_size": 151665
|
|
}
|
|
|