GSM8K
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
GSM8K numeric-remapping (attack set)
87.84Accuracy
3
GSM8K baseline original
1,885Correct Count
3
GSM8K
95.15Pass@8
3
GSM8K
81.79Mean@8 Accuracy
3
GSM8K
53.9GSM8K Accuracy
3
GSM8K (test)
65.4Throughput (Tokens/sec)
3
GSM8K (held-out)
4FPR
3
GSM8K (test)
93.86Accuracy
3
GSM8K
38.89EM (Strict)
3
GSM8K
87.22Teacher Accuracy
3
GSM8K
0Risk
3
GSM8K (eval)
0.3Risk
3
GSM8K (test)
99Accuracy
3
GSM8K
76.58Accuracy
3
GSM8K
138Throughput (tok/s)
3
GSM8K
95.38Accuracy (mean)
3
GSM8K (test)
37.38Accuracy (Ckpt 1)
3
GSM8K
89.16Accuracy
3
GSM8K
6.5Total Tokens Processed
3
GSM8k
58.5Performance
3
GSM8K HARD (test)
37.08Accuracy
3
GSM8K Right Dense variant (test)
75Success Rate
3
GSM8K Mid Dense variant (test)
68Success Rate
3
GSM8K Left Dense variant (test)
77Success Rate
3
GSM8K
93.07Mean Accuracy (±STD)
3