Skip to content

Long Horizon dataset #65

@maitreygram

Description

@maitreygram

Thank you for releasing this dataset and congratulations on the paper! In your paper, you mention that you use long horizon prompts to collect the dataset. I was wondering if you also release the long horizon instructions (e.g. “make a smiley face out of blocks” etc.), along with the corresponding short horizon instruction sequence and the trajectories, and if you could point me to the dataset if it is available?

From the dataset description, I can only find these datasets -

Dataset Real/sim Controlled by Language-labeled by # episodes
language_table real human human 442,226
language_table_sim sim human human 181,020
language_table_blocktoblock_sim sim human scripted 8,000
language_table_blocktoblock_4block_sim sim human scripted 8,298
language_table_blocktoblock_oracle_sim sim oracle scripted 200,000
language_table_blocktoblockrelative_oracle_sim sim oracle scripted 200,000
language_table_blocktoabsolute_oracle_sim sim oracle scripted 200,000
language_table_blocktorelative_oracle_sim sim oracle scripted 200,000
language_table_separate_oracle_sim sim oracle scripted 200,000

and it looks like these are all short horizon datasets. Are there additional links to download the long horizon goals and the trajectories associated with them?

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions