Mathematical Reasoning on GSM8K (test) (Accuracy, AVG, MAX)
32AccuracyDrICL
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DrICLBackbone=Llama-2-7b-chat-hf, k-shot=102025.01 | 32 | 29 | 32 | |
| ITBackbone=Llama-2-7b-chat-hf, k-shot=02025.01 | 31 | 24 | 31 | |
| DrICLBackbone=Llama-2-7b-chat-hf, k-shot=32025.01 | 31 | 29 | 32 | |
| DrICLBackbone=Llama-2-7b-chat-hf, k-shot=02025.01 | 30 | 29 | 32 | |
| NFTBackbone=Llama-2-7b-chat-hf, k-shot=02025.01 | 28 | 11 | 28 | |
| DrICLBackbone=Llama-2-7b-chat-hf, k-shot=12025.01 | 28 | 29 | 32 | |
| DrICLBackbone=Llama-2-7b-chat-hf, k-shot=52025.01 | 27 | 29 | 32 | |
| ITBackbone=Llama-2-7b-chat-hf, k-shot=12025.01 | 26 | 24 | 31 | |
| ITBackbone=Llama-2-7b-chat-hf, k-shot=32025.01 | 26 | 24 | 31 | |
| ITBackbone=Llama-2-7b-chat-hf, k-shot=102025.01 | 26 | 24 | 31 | |
| MetaICLBackbone=Llama-2-7b-chat-hf, k-shot=12025.01 | 26 | 25 | 26 | |
| MetaICLBackbone=Llama-2-7b-chat-hf, k-shot=32025.01 | 26 | 25 | 26 | |
| MetaICLBackbone=Llama-2-7b-chat-hf, k-shot=202025.01 | 26 | 25 | 26 | |
| DrICLBackbone=Llama-2-7b-chat-hf, k-shot=202025.01 | 26 | 29 | 32 | |
| MetaICLBackbone=Llama-2-7b-chat-hf, k-shot=02025.01 | 24 | 25 | 26 | |
| MetaICLBackbone=Llama-2-7b-chat-hf, k-shot=52025.01 | 24 | 25 | 26 | |
| MetaICLBackbone=Llama-2-7b-chat-hf, k-shot=102025.01 | 24 | 25 | 26 | |
| ITBackbone=Llama-2-7b-chat-hf, k-shot=52025.01 | 21 | 24 | 31 | |
| NFTBackbone=Llama-2-7b-chat-hf, k-shot=12025.01 | 16 | 11 | 28 | |
| ITBackbone=Llama-2-7b-chat-hf, k-shot=202025.01 | 16 | 24 | 31 | |
| NFTBackbone=Llama-2-7b-chat-hf, k-shot=32025.01 | 11 | 11 | 28 | |
| NFTBackbone=Llama-2-7b-chat-hf, k-shot=52025.01 | 7 | 11 | 28 | |
| NFTBackbone=Llama-2-7b-chat-hf, k-shot=202025.01 | 2 | 11 | 28 | |
| NFTBackbone=Llama-2-7b-chat-hf, k-shot=102025.01 | 1 | 11 | 28 |