QwQ-14B-Math-v0.2 / config.json
qingy2024's picture
Trained with Unsloth
f3654eb verified
raw
history blame contribute delete
822 Bytes
{
"_name_or_path": "unsloth/qwen2.5-14b-bnb-4bit",
"architectures": [
"Qwen2ForCausalLM"
],
"attention_dropout": 0.0,
"bos_token_id": 151643,
"eos_token_id": 151643,
"hidden_act": "silu",
"hidden_size": 5120,
"initializer_range": 0.02,
"intermediate_size": 13824,
"max_position_embeddings": 131072,
"max_window_layers": 48,
"model_type": "qwen2",
"num_attention_heads": 40,
"num_hidden_layers": 48,
"num_key_value_heads": 8,
"pad_token_id": 151665,
"rms_norm_eps": 1e-05,
"rope_scaling": null,
"rope_theta": 1000000.0,
"sliding_window": null,
"tie_word_embeddings": false,
"torch_dtype": "bfloat16",
"transformers_version": "4.47.1",
"unsloth_fixed": true,
"unsloth_version": "2024.12.4",
"use_cache": true,
"use_sliding_window": false,
"vocab_size": 152064
}