Datasets:

---

dancher00

/

pickcube-v1-smolvla

like 0

Hey, I have a question Please, you used "language_instruction" in each row, but original dataset smol vla was trained on didn't use it (And i see you are finetuning not training from scratch), did you change anything in the model? did the model understood the instruction given language instruction?