LAION-AI / LAION-AI/Open-Assistant

What is the details for the data used for reward model OpenAssistant/oasst-rm-2-pythia-6.9b-epoch-1?

Open
#3,686 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
37.4k
Forks
3.3k
PR merge metrics
No merged PRs in 30d

Description

Many thanks for your great open sourcing effort! And I am new to this field and I am particularly interested in the data for training reward model. I noticed that there is a simple dataset config for this, but I am a little bit confused about the details.

  datasets:
    - oasst_export:
        lang: "en,es,de,fr"
        input_file_path: 2023-03-27_oasst_research_ready_synth.jsonl.gz
        val_split: 0.1
    - anthropic_rlhf:
        fraction: 0.1
        max_val_set: 1000
    - shp:
        max_val_set: 1000
    - hellaswag:
        fraction: 0.5
        max_val_set: 1000
    - webgpt:
         val_split: 0.05
         max_val_set: 1000
    - hf_summary_pairs:
         fraction: 0.1
         max_val_set: 250

How can we use hellaswag as a comparison dataset? There seem to be multiple choices(rather than 2)
Is there any experimental evidence support the fraction setting we are currently using?

Many thanks for any responses in advance!

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reviewing the reward-model dataset configuration shown in the issue and the definitions of its listed datasets, especially hellaswag and the fraction settings. Document the data sources, how multi-choice examples are used for comparison, and any experimental basis for the configured fractions.

Written by the indexing model from the issue text.

Assessment

Domain
data, machine-learning
Issue type
Documentation
Difficulty
5/5
Estimated time
Over a week
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
20/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.