nynorsk_second_test_GRPO / generation_config.json
pere's picture
GRPO model (assistant split heuristic reward)
78decbe verified
raw
history blame contribute delete
241 Bytes
{
"_from_model_config": true,
"bos_token_id": 128000,
"do_sample": true,
"eos_token_id": [
128009
],
"max_new_tokens": 768,
"pad_token_id": 128001,
"temperature": 0.7,
"top_p": 0.9,
"transformers_version": "4.52.4"
}