Code Generation on HumanEval (val)
490Pass@1DeepSeekMoE
Evaluation Results
| Method | Links | |
|---|---|---|
| DeepSeekMoE# Shot=0-shot, # Total Params=2.0B, # Activated Params=0.3B, FLOPs per 2K Tokens=4.3T, # Training Tokens=100B2024.01 | 490 | |
| GShard# Shot=0-shot, # Total Params=2.0B, # Activated Params=0.3B, FLOPs per 2K Tokens=4.3T, # Training Tokens=100B2024.01 | 370 | |
| Switch# Shot=0-shot, # Total Params=2.0B, # Activated Params=0.2B, FLOPs per 2K Tokens=2.9T, # Training Tokens=100B2024.01 | 240 | |
| Hash Layer# Shot=0-shot, # Total Params=2.0B, # Activated Params=0.2B, FLOPs per 2K Tokens=2.9T, # Training Tokens=100B2024.01 | 120 | |
| Dense# Shot=0-shot, # Total Params=0.2B, # Activated Params=0.2B, FLOPs per 2K Tokens=2.9T, # Training Tokens=100B2024.01 | 0 |