Graph Node Classification and Link Prediction on HuggingFace
78.76n-F1GraphSAGE
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| GraphSAGELLM=GPT-4-turbo2024.05 | 78.76 | 52.53 | 34.09 | |
| GraphSAGEco-trainLLM=GPT-4-turbo2024.05 | 78.49 | 52.62 | 33.88 | |
| GraphSAGELLM=GPT-3.5-turbo2024.05 | 77.9 | 52.68 | 35.11 | |
| GraphSAGEco-trainLLM=GPT-3.5-turbo2024.05 | 77.87 | 53.04 | 35.32 | |
| DirectLLM=GPT-4-turbo2024.05 | 77.6 | 52.18 | 33.68 | |
| DirectLLM=GPT-3.5-turbo2024.05 | 73.85 | 45.73 | 28.95 | |
| GraphSAGEco-trainLLM=CodeLlama-13B2024.05 | 68.92 | 44.85 | 29.58 | |
| TAPELLM=GPT-4-turbo2024.05 | 68.82 | — | 9.5 | |
| GraphSAGELLM=Mistral-7B2024.05 | 68.12 | 43.09 | 25.77 | |
| TAPELLM=GPT-3.5-turbo2024.05 | 68 | — | 8.83 | |
| GraphSAGEco-trainLLM=Mistral-7B2024.05 | 67.61 | 43.14 | 27.2 | |
| GraphSAGELLM=CodeLlama-13B2024.05 | 67.3 | 42.41 | 26.56 | |
| GraphTokenLLM=Mistral-7B2024.05 | 64.42 | 32.04 | 18.6 | |
| TAPELLM=CodeLlama-13B2024.05 | 64.03 | — | 8.05 | |
| GraphTokenLLM=Vicuna-13B2024.05 | 63.37 | 31.54 | 15.61 | |
| GraphSAGEco-trainLLM=Vicuna-13B2024.05 | 62.82 | 37.04 | 19.68 | |
| GraphTokenLLM=CodeLlama-13B2024.05 | 62.15 | 32.55 | 20.08 | |
| GraphSAGELLM=Vicuna-13B2024.05 | 61.86 | 35.68 | 20.08 | |
| TAPELLM=Mistral-7B2024.05 | 61.82 | — | 6.13 | |
| DirectLLM=Mistral-7B2024.05 | 60.6 | 30.23 | 16.36 | |
| TAPELLM=Vicuna-13B2024.05 | 59.47 | — | 5.07 | |
| DirectLLM=CodeLlama-13B2024.05 | 57.55 | 28.88 | 14.29 | |
| DirectLLM=Vicuna-13B2024.05 | 50.46 | 21.27 | 8.72 |