Inference Memory on inference memory
40Success RatePrior Knowledge (Bench)
Evaluation results
| Method | Links | |
|---|---|---|
| Prior Knowledge (Bench)Target Model=Mistral 0.3-7B2025.12 | 40 | |
| Prior Knowledge (Bench)Target Model=Llama 3.1-8B2025.12 | 40 | |
| Prior Knowledge (Bench)Target Model=Qwen 3-8B2025.12 | 35 | |
| EMBER (Multi)Target Model=Mistral 0.3-7B2025.12 | 25 | |
| EMBER (Multi)Target Model=Llama 3.1-8B2025.12 | 22.2 | |
| EMBER (Multi)Target Model=Qwen 3-8B2025.12 | 20 | |
| EMBER (Single)Target Model=Llama 3.1-8B2025.12 | 18.7 | |
| EMBER (Single)Target Model=Mistral 0.3-7B2025.12 | 16.2 | |
| EMBER (Multi)Target Model=Qwen 3-32B2025.12 | 16.2 | |
| EMBER (Single)Target Model=Qwen 3-8B2025.12 | 15.3 | |
| EMBER (Multi)Target Model=Qwen 3-14B2025.12 | 13.2 | |
| EMBER (Single)Target Model=Qwen 3-32B2025.12 | 11.8 | |
| EMBER (Single)Target Model=Qwen 3-14B2025.12 | 10.8 | |
| Prior Knowledge (Prompt)Target Model=Mistral 0.3-7B2025.12 | 10 | |
| Prior Knowledge (Prompt)Target Model=Qwen 3-8B2025.12 | 8.5 | |
| Prior Knowledge (Prompt)Target Model=Qwen 3-14B2025.12 | 7.5 | |
| Prior Knowledge (Prompt)Target Model=Qwen 3-32B2025.12 | 6.5 | |
| Prior Knowledge (Prompt)Target Model=Llama 3.1-8B2025.12 | 6 | |
| Prior Knowledge (Bench)Target Model=Qwen 3-14B2025.12 | 5 | |
| Prior Knowledge (Bench)Target Model=Qwen 3-32B2025.12 | 5 | |
| SFTTarget Model=Mistral 0.3-7B2025.12 | 0 | |
| SFTTarget Model=Llama 3.1-8B2025.12 | 0 | |
| SFTTarget Model=Qwen 3-8B2025.12 | 0 | |
| SFTTarget Model=Qwen 3-14B2025.12 | 0 | |
| SFTTarget Model=Qwen 3-32B2025.12 | 0 |