quasar-phase3-kl2-27 / generation_config.json
algo-dev99's picture
Initial Quasar 3B FKL-pretrain checkpoint, sparse top-128 KL=2.27
860729d verified
Raw
History Blame Contribute Delete
166 Bytes
{
"_from_model_config": true,
"bos_token_id": 1,
"do_sample": true,
"eos_token_id": 2,
"top_k": 4,
"transformers_version": "5.7.0",
"use_cache": true
}