75640a279f
- bec5745161bec60d1334228c0d9eb0758fa9e879bdaa006fc37b8f75e062fc13 (5af833d4c6bb1381fd031dbe2b4987502f710938) - f81eed435216eae10b326c39fb92406602efa2e80318342f3ad4cc671c8d277c (c657af2ec95b0679732ca1b91aa42d8d94978bda) - 8df993c820a76bf3395e3694c09712af8d54be13ab318d652c2cb19a23c3f941 (f6daea9a6f54df64777c7c18678e546c80075e4e) - 29ca5bff158c3065cab105a21a9509e59b19aafce4844b3d9731bad4b9fea69e (bbba15738490e051f00d3a4b5ab62130458d5e46) - 366858fe0ad9652394d140fd2122ec792234a42d9d97fba688480459d0b0eea3 (6d44f36d54bc6b14d44317bacee0a413ed70e70b)
35 lines
826 B
JSON
35 lines
826 B
JSON
{
|
|
"_name_or_path": "/sailor2/Sailor2_Release/Sailor2-8B",
|
|
"architectures": [
|
|
"Qwen2ForCausalLM"
|
|
],
|
|
"attention_dropout": 0.0,
|
|
"bos_token_id": 151643,
|
|
"eos_token_id": 151643,
|
|
"hidden_act": "silu",
|
|
"hidden_size": 3584,
|
|
"initializer_range": 0.02,
|
|
"intermediate_size": 18944,
|
|
"max_position_embeddings": 131072,
|
|
"max_window_layers": 28,
|
|
"model_type": "qwen2",
|
|
"newpos": true,
|
|
"num_attention_heads": 28,
|
|
"num_hidden_layers": 32,
|
|
"num_key_value_heads": 4,
|
|
"rms_norm_eps": 1e-06,
|
|
"rope_scaling": {
|
|
"factor": 4.0,
|
|
"type": "default"
|
|
},
|
|
"rope_theta": 1000000.0,
|
|
"sliding_window": null,
|
|
"tie_word_embeddings": false,
|
|
"torch_dtype": "bfloat16",
|
|
"transformers_version": "4.39.1",
|
|
"use_cache": true,
|
|
"use_mrope": false,
|
|
"use_sliding_window": false,
|
|
"vocab_size": 151936
|
|
}
|