pbsftmix-orig-safety60-nosys-normal-3b / generation_config.json
Raghav-Singhal's picture
instruct+safety mix SFT (field=messages_original, 60% safety = 180000 safety + 120000 instruct of 300000; instruct=jkminder/model-raising-pbsft-instruct-300k safety=jkminder/model-raising-pbsft-safety-180k, template=epe-template-nosys, tokenizer=/capstor/store/cscs/swissai/a141/model-raising-training/checkpoints/pretraining/smollm2-3b/hf/epe-1p-3b-llama3arch-smollm2tok-500B-40n-2048sl-960gbsz-no_bce) on normal-3b-llama3arch-smollm2tok-500B-40n-2048sl-960gbsz
0a3e084 verified
Raw
History Blame Contribute Delete
171 Bytes
{
"bos_token_id": 1,
"do_sample": true,
"eos_token_id": [
2
],
"pad_token_id": 2,
"temperature": 0.6,
"top_p": 0.9,
"transformers_version": "4.57.6"
}