Triplet Alignment on AG-News
85.9Triplet Alignment Accuracy (AG-News)LoRA
Evaluation results
| Method | Links | ||
|---|---|---|---|
| LoRABackbone=Qwen2.5-7B, Evaluation Protocol=Fine-tuning2026.05 | 85.9 | — | |
| PromptEmbedderBackbone=Qwen2.5-7B, Evaluation Protocol=Frozen Backbone Methods2026.05 | 85.36 | — | |
| LoRABackbone=Mistral-7B, Evaluation Protocol=Fine-tuning2026.05 | 84.81 | — | |
| PromptEmbedderBackbone=LLaMA-3.2-3B, Evaluation Protocol=Frozen Backbone Methods2026.05 | 84.17 | — | |
| PromptEmbedderBackbone=LLaMA-3.2-1B, Evaluation Protocol=Frozen Backbone Methods2026.05 | 83.62 | — | |
| PromptEmbedderBackbone=Mistral-7B, Evaluation Protocol=Frozen Backbone Methods2026.05 | 82.24 | — | |
| Prompt-TuningBackbone=Qwen2.5-7B, Evaluation Protocol=Frozen Backbone Methods2026.05 | 81.91 | — | |
| Prompt-TuningBackbone=Mistral-7B, Evaluation Protocol=Frozen Backbone Methods2026.05 | 81.6 | — | |
| LoRABackbone=LLaMA-3.2-3B, Evaluation Protocol=Fine-tuning2026.05 | 80.92 | — | |
| Prompt-TuningBackbone=LLaMA-3.2-3B, Evaluation Protocol=Frozen Backbone Methods2026.05 | 80.62 | — | |
| LoRABackbone=LLaMA-3.2-1B, Evaluation Protocol=Fine-tuning2026.05 | 80.37 | — | |
| PromptEOLBackbone=Qwen2.5-7B, Evaluation Protocol=Training-free Prompting2026.05 | 78.57 | — | |
| EchoBackbone=Qwen2.5-7B, Evaluation Protocol=Training-free Prompting2026.05 | 78.54 | — | |
| Pretend-CoTBackbone=Qwen2.5-7B, Evaluation Protocol=Training-free Prompting2026.05 | 76.77 | — | |
| EchoBackbone=LLaMA-3.2-1B, Evaluation Protocol=Training-free Prompting2026.05 | 76.13 | — | |
| EchoBackbone=Mistral-7B, Evaluation Protocol=Training-free Prompting2026.05 | 76.11 | — | |
| EchoBackbone=LLaMA-3.2-3B, Evaluation Protocol=Training-free Prompting2026.05 | 75.74 | — | |
| Pretend-CoTBackbone=LLaMA-3.2-3B, Evaluation Protocol=Training-free Prompting2026.05 | 72.87 | — | |
| Prompt-TuningBackbone=LLaMA-3.2-1B, Evaluation Protocol=Frozen Backbone Methods2026.05 | 72.8 | — | |
| PromptEOLBackbone=Mistral-7B, Evaluation Protocol=Training-free Prompting2026.05 | 72.63 | — | |
| Pretend-CoTBackbone=Mistral-7B, Evaluation Protocol=Training-free Prompting2026.05 | 71.81 | — | |
| PromptEOLBackbone=LLaMA-3.2-3B, Evaluation Protocol=Training-free Prompting2026.05 | 71 | — | |
| PromptEOLBackbone=LLaMA-3.2-1B, Evaluation Protocol=Training-free Prompting2026.05 | 68.1 | — | |
| Pretend-CoTBackbone=LLaMA-3.2-1B, Evaluation Protocol=Training-free Prompting2026.05 | 66.21 | — | |
| BGE2025.05 | — | 71.46 | |
| GSTransformBackbone=UAE2025.05 | — | 87.13 | |
| GSTransformBackbone=Mxbai2025.05 | — | 87.27 | |
| GSTransformBackbone=BGE2025.05 | — | 87.01 | |
| InBedderBackbone=Roberta2025.05 | — | 80.99 | |
| InBedderBackbone=Llama22025.05 | — | 86.93 | |
| InstructOR2025.05 | — | 73.41 | |
| Mxbai2025.05 | — | 75.41 | |
| UAE2025.05 | — | 72.59 |