Step recognition on COIN
67.3Top-1 AccuracyGAD
Evaluation results
| Method | Links | |
|---|---|---|
| GADBackbone=Llama3-8B-Instruct, Classifier Type=Generative-Assisted Discriminative2026.03 | 67.3 | |
| GADEvaluation Protocol=fine-tuned2026.03 | 67.3 | |
| DiscBackbone=Llama3-8B-Instruct, Classifier Type=Discriminative2026.03 | 66.4 | |
| GADBackbone=Llama3.2-1B-Instruct, Classifier Type=Generative-Assisted Discriminative2026.03 | 65.3 | |
| DiscBackbone=Llama3.2-1B-Instruct, Classifier Type=Discriminative2026.03 | 64.1 | |
| StreamMind-8BBackbone=8B2026.03 | 63.7 | |
| Videollm-MoD-8BBackbone=8B2026.03 | 63.4 | |
| Videollm-online-8BBackbone=8B2026.03 | 63.1 | |
| VideoTaskGraph2026.03 | 57.2 | |
| Qwen2.5-VL-7BEvaluation Protocol=open-ended2026.03 | 16.1 | |
| Qwen2.5-VL-7BPrompting Strategy=categories in prompt2026.03 | 11.9 | |
| Videollm-online-8BEvaluation Protocol=open-ended2026.03 | 4.8 |