Instruction Following on Real-world Robot Manipulation Generalization v1 (test)
80Success RateM2-VLA
Evaluation results
| Method | Links | |
|---|---|---|
| M2-VLAVLM Fine-tuning=No2026.04 | 80 | |
| VLA-AdapterVLM Fine-tuning=Yes2026.04 | 55 | |
| smolVLAVLM Fine-tuning=No2026.04 | 25 | |
| VLA-Adapter (Frozen)VLM Fine-tuning=No2026.04 | 5 |