ShuaiYang03/instructvla_pretraining_v2_query_64_lora_state

Model Card for InstructVLA-SimplerEnv (with state)

Weights

• checkpoints/step-055500-epoch-08-loss=0.0819.pt: Complete model checkpoint for direct evaluation.
• checkpoints/action_model_55500_unload.pt: Model containing only the VLM backbone and action expert. This cannot be used independently; it is intended solely for initializing Stage-2 training.
• checkpoints/action_model_55500: Action LoRA adapter. This cannot be used independently; it is intended solely for initializing Stage-2 training.

Evaluation Results

• results_step-055500-epoch-08-loss=0.0819_instruct_1-3: Results on SimplerEnv-Instruct.
• results_step-055500-epoch-08-loss=0.0819_1-3: Results on SimplerEnv.