gemma-3-1b-it-Math-SFT / generation_config.json
NotoriousH2's picture
SFT only (teacher distillation from Qwen3-30B). GSM8K avg ~44.9%
6959d46 verified
raw
history blame contribute delete
217 Bytes
{
"bos_token_id": 2,
"cache_implementation": "hybrid",
"do_sample": true,
"eos_token_id": [
1,
1,
106
],
"pad_token_id": 1,
"top_k": 64,
"top_p": 0.95,
"transformers_version": "4.57.3"
}