24 lines
537 B
JSON
24 lines
537 B
JSON
|
{
|
||
|
|
"architectures": [
|
||
|
|
"Qwen3ForCausalLM"
|
||
|
|
],
|
||
|
|
"model_type": "qwen3",
|
||
|
|
"vocab_size": 65536,
|
||
|
|
"hidden_size": 1280,
|
||
|
|
"intermediate_size": 3456,
|
||
|
|
"num_hidden_layers": 24,
|
||
|
|
"num_attention_heads": 20,
|
||
|
|
"num_key_value_heads": 5,
|
||
|
|
"head_dim": 64,
|
||
|
|
"max_position_embeddings": 4096,
|
||
|
|
"rms_norm_eps": 1e-06,
|
||
|
|
"rope_theta": 500000.0,
|
||
|
|
"attention_bias": false,
|
||
|
|
"tie_word_embeddings": true,
|
||
|
|
"hidden_act": "silu",
|
||
|
|
"torch_dtype": "bfloat16",
|
||
|
|
"bos_token_id": null,
|
||
|
|
"eos_token_id": 0,
|
||
|
|
"pad_token_id": 1,
|
||
|
|
"use_cache": true
|
||
|
|
}
|