MiniMind2-SFT-104M / generation_config.json
leixinlin's picture
MiniMind2-SFT-104M: 4B pretrain + 4.8B SFT tokens, Llama format
a0ca88c verified
Raw
History Blame Contribute Delete
194 Bytes
{
"_from_model_config": true,
"bos_token_id": 1,
"eos_token_id": 2,
"output_attentions": false,
"output_hidden_states": false,
"transformers_version": "5.4.0",
"use_cache": true
}