Non-thinking version of cs-giung/gsm8k-aug-filtered-qwen3-0.6b-2048.
Responses were generated with Qwen/Qwen3-0.6B using the Qwen3 chat template and
the non-thinking generation settings from the official Qwen3-0.6B release:
temperature = 0.7
top_p = 0.8
top_k = 20
max_new_tokens = 2048
Thinking mode was disabled through the Qwen3 chat template during generation.
For instances exhibiting significant… See the full description on the dataset page:
https://huggingface.co/datasets/cs-giung/gsm8k-aug-filtered-qwen3-0.6b-no-think-2048.
