Libero_90 "LIVING_ROOM_SCENE1_pick_up_the_tomato_sauce_and_put_it_in_the_basket_demo.hdf5" file truncated error

[Zzfiv3]

Hi!
When I want to read the file "LIVING_ROOM_SCENE1_pick_up_the_tomato_sauce_and_put_it_in_the_basket_demo.hdf5" in Libero_90 datasets, the code is shown below:

print(f"Processing {hdf5_path}")
        try:
            hdf5_file = h5py.File(os.path.join(base_dir, hdf5_path), 'r', swmr=False, libver='latest')
            demos = list(hdf5_file["data"].keys())
        except Exception as e:
            print(f"Error processing {hdf5_path}: {e}")
            continue

I found the output has a truncated error:

Processing LIVING_ROOM_SCENE1_pick_up_the_tomato_sauce_and_put_it_in_the_basket_demo.hdf5
Error processing LIVING_ROOM_SCENE1_pick_up_the_tomato_sauce_and_put_it_in_the_basket_demo.hdf5: Unable to synchronously open file (truncated file: eof = 590254080, sblock->base_addr = 0, stored_eof = 806284784)

Is anyone in the same situation as me, help!

[Kai-Sheng-caesar]

Hi, I have the same problem, do you have any solution?

[Zzfiv3]

@Kai-Sheng-caesar Yes! I found that if you download the dataset from the huggingface (using python benchmark_scripts/download_libero_datasets.py --use-huggingface or python benchmark_scripts/download_libero_datasets.py), one subset of the dataset is broken. The size of the dataset should be around 800Mb, but the huggingface only has around 300Mb. But it seems they have fixed it around 15 hours ago.

[Kai-Sheng-caesar]

OK，thank you very much!