NickyNicky/

Llama-1B-base-GRPO-miniThinky_v0

dataset: NickyNicky/ngxson_MiniThinky_v1_deduplicated_11_percent
** full train
** 360 row
** 11 epoch
** max token 512
** time: 2:38:24

<reasoning>
...
</reasoning>
<answer>
...
</answer>