Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Mathematical Reasoning on GSM8K (Pass@1, Token count)

95Pass@1 Accuracy

RED

89.38490.84292.393.758Apr 3, 2026
Updated 12d ago

Evaluation Results

MethodLinks
2026.04
95465
2026.04
94.9269
2026.04
94.7822
2026.04
94.7239
2026.04
94.7269
2026.04
94.7473
2026.04
94.6775
2026.04
94.6809
2026.04
94.6443
2026.04
94.6209
2026.04
94.51,095
2026.04
94.5641
2026.04
94.51,558
2026.04
94.5249
2026.04
94.41,373
2026.04
94.41,090
2026.04
94.4420
2026.04
94.4255
2026.04
94.3711
2026.04
94.31,435
2026.04
94.2599
2026.04
94.2609
2026.04
94.2427
2026.04
94.11,573
2026.04
94.1483
2026.04
94.1823
2026.04
94.1792
2026.04
94.1271
2026.04
94.1433
2026.04
94432
2026.04
93.9438
2026.04
93.9266
2026.04
93.9239
2026.04
93.7297
2026.04
93.7241
2026.04
93.4459
2026.04
93.3247
2026.04
93.3277
2026.04
93.2432
2026.04
93.1598
2026.04
93.1283
2026.04
92.9264
2026.04
92.9571
2026.04
92.5291
2026.04
92.5290
2026.04
92.5257
2026.04
92.4442
2026.04
92.3606
2026.04
92.3446
2026.04
92.1275
2026.04
91.9388
2026.04
91.8301
2026.04
89.7297
2026.04
89.6394