AmareshHebbar/ShortThink_Traces_Qwen3_1.7B
Viewer • Updated • 5.13k
Qwen3 1.7B reasoning models trained to stop overthinking. Same data, six algorithms, one honest comparison table.
Note Training traces shared by every model in this collection
Note ShortThink SFT
Note ShortThink DPO
Note ShortThink GRPO
Note ShortThink TIES