suryadv/qwen3-8b-aime-2009-2024-16x
Viewer • Updated • 7.68k • 1.39k
IncorrectTraceSFT
Math models and data for How Should Incorrect Traces Be Used in Supervised Fine-Tuning?
-
-
suryadv/IncorrectTraceSFT-Qwen2.5-7B-AIME2024-correct
Text Generation • 8B • Updated • 252 -
suryadv/IncorrectTraceSFT-Qwen2.5-7B-AIME2024-incorrect
Text Generation • 8B • Updated • 239 -
suryadv/IncorrectTraceSFT-Qwen2.5-7B-AIME2025-correct
Text Generation • 8B • Updated • 232 -
suryadv/IncorrectTraceSFT-Qwen2.5-7B-AIME2025-incorrect
Text Generation • 8B • Updated • 236 -
suryadv/IncorrectTraceSFT-Qwen2.5-7B-AIME-SplitA-correct
Text Generation • 8B • Updated • 240 -
suryadv/IncorrectTraceSFT-Qwen2.5-7B-AIME-SplitA-incorrect
Text Generation • 8B • Updated • 227 -
suryadv/IncorrectTraceSFT-Qwen2.5-7B-AIME-SplitB-correct
Text Generation • 8B • Updated • 244 -
suryadv/IncorrectTraceSFT-Qwen2.5-7B-AIME-SplitB-incorrect
Text Generation • 8B • Updated • 247 -
suryadv/IncorrectTraceSFT-Qwen2.5-7B-MATH-unmasked
Text Generation • 8B • Updated • 257 -
suryadv/IncorrectTraceSFT-Qwen2.5-7B-MATH-random-mask-25
Text Generation • 8B • Updated • 260 -
suryadv/IncorrectTraceSFT-Qwen2.5-7B-MATH-random-mask-50
Text Generation • 8B • Updated • 251 -
suryadv/IncorrectTraceSFT-Qwen2.5-7B-MATH-random-mask-70
Text Generation • 8B • Updated • 250 -
suryadv/IncorrectTraceSFT-Qwen2.5-7B-MATH-random-mask-80
Text Generation • 8B • Updated • 253 -
suryadv/IncorrectTraceSFT-Qwen2.5-7B-MATH-random-mask-90
Text Generation • 8B • Updated • 268 -
suryadv/IncorrectTraceSFT-Qwen2.5-7B-MATH-correct-only
Text Generation • 8B • Updated • 261