generation_config.json
312 B · 17 lines · json Raw
1 {
2 "attn_implementation": "flash_attention_2",
3 "bos_token_id": 151643,
4 "do_sample": true,
5 "eos_token_id": [
6 151645,
7 151643
8 ],
9 "pad_token_id": 151643,
10 "repetition_penalty": 1.05,
11 "temperature": 0.1,
12 "top_k": 1,
13 "top_p": 0.001,
14 "transformers_version": "4.49.0",
15 "use_cache": false
16 }
17