target-matched
kimi-linear-48b-a3b-target-matched-math-240k
kimi-linear-48b-a3b-target-matched-math-240k
239,467 rows of math-reasoning trajectories regenerated against
moonshotai/Kimi-Linear-48B-A3B-Instruct as the target model. Used to train DFlash
speculative-decoding drafters in
la-draftery.
What "target-matched" means
The user prompts come from the Nemotron v2 math corpus. The assistant
completions in this dataset are the target model's own outputs — each
prompt was sent to moonshotai/Kimi-Linear-48B-A3B-Instruct and its… See the full description on the dataset page: https://huggingface.co/datasets/Moonlight556/kimi-linear-48b-a3b-target-matched-math-240k.qwen3.5-0.8b-target-matched-math-240k
qwen3.5-0.8b-target-matched-math-240k
239,467 rows of math-reasoning trajectories regenerated against
Qwen/Qwen3.5-0.8B as the target model. Used to train DFlash
speculative-decoding drafters in
la-draftery.
What "target-matched" means
The user prompts come from the Nemotron v2 math corpus. The assistant
completions in this dataset are the target model's own outputs — each
prompt was sent to Qwen/Qwen3.5-0.8B and its completion was captured.
Drafters trained on… See the full description on the dataset page: https://huggingface.co/datasets/Moonlight556/qwen3.5-0.8b-target-matched-math-240k.