I'm running the 04_eval_finetuned_on_robot.py script provided in examples, straight out of the box. I tried using the language conditioning option, and providing a new instruction of "move the green object to the bottom right burner" when prompted.
Running into this issue in the pass through the model - wondering if this is a common problem or if there's any hints on fixing this?
File "/home/arhan/projects/widowx_octo_inference/04_eval_finetuned_on_robot.py", line 238, in <module>
app.run(main)
File "/home/arhan/miniconda3/envs/octo/lib/python3.10/site-packages/absl/app.py", line 308, in run
_run_main(main, args)
File "/home/arhan/miniconda3/envs/octo/lib/python3.10/site-packages/absl/app.py", line 254, in _run_main
sys.exit(main(argv))
File "/home/arhan/projects/widowx_octo_inference/04_eval_finetuned_on_robot.py", line 212, in main
action = np.array(policy_fn(obs, task), dtype=np.float64)
File "/home/arhan/projects/widowx_octo_inference/04_eval_finetuned_on_robot.py", line 127, in sample_actions
actions = pretrained_model.sample_actions(
File "/home/arhan/projects/octo/octo/model/octo_model.py", line 187, in sample_actions
transformer_outputs = self.run_transformer(
File "/home/arhan/projects/octo/octo/model/octo_model.py", line 152, in run_transformer
return self.module.apply(
File "/home/arhan/projects/octo/octo/model/octo_module.py", line 274, in __call__
outputs["obs"] = TokenGroup.concatenate(
File "/home/arhan/projects/octo/octo/model/components/base.py", line 31, in concatenate
data = jnp.concatenate([t.tokens for t in group_list], axis=axis)
File "/home/arhan/miniconda3/envs/octo/lib/python3.10/site-packages/jax/_src/numpy/lax_numpy.py", line 1839, in concatenate
raise ValueError("Need at least one array to concatenate.")
ValueError: Need at least one array to concatenate.```
I'm running the 04_eval_finetuned_on_robot.py script provided in examples, straight out of the box. I tried using the language conditioning option, and providing a new instruction of "move the green object to the bottom right burner" when prompted.
Running into this issue in the pass through the model - wondering if this is a common problem or if there's any hints on fixing this?