GSM8K
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
GSM8K
51.1GSM8K Accuracy
7
GSM8K
0.815Acc
7
GSM8K
76.7ASR
7
GSM8K 3-shot
92.27Accuracy
7
GSM8k
96.2Accuracy (Baseline)
7
GSM8K (eval)
42.29MAE
6
GSM8K LLaMA-2-7B (test)
66.5Accuracy
6
GSM8K 1319 problems (test)
96.59Accuracy
6
GSM8K
0.706Reconstruction Rate
6
GSM8K (test)
59Accuracy
6
GSM8K 500 samples for LLaMA, 50 for Qwen and DeepSeek (test)
2,578.42Avg Tokens
6
GSM8K 50 randomly sampled questions (test)
48.57PCR
6
GSM8K
58.82PCR
6
GSM8K
95.53Accuracy
6
GSM8K (N = 100)
0FPR
6
GSM8K 5-shot v1.0 (test)
73.7Accuracy
6
GSM8K
78.9Strict-match Accuracy
6
GSM8K (test)
84.04AUROC
6
GSM8K Single Client
74.3Accuracy
6
GSM8K (test)
96.13Accuracy (Raw Consensus)
6
GSM8K (test)
70Accuracy (Error <= 1)
6
GSM8K (test)
1.13MAE (Scale 1)
6
GSM8k
93.9Pass@1
6
GSM8K
1.741Decode Speed (tokens/s)
6
GSM8K
94.14Accuracy (%)
6