Skip to content

[pull] master from microsoft:master#58

Merged
pull[bot] merged 3 commits intospecialized806:masterfrom
deepspeedai:master
Nov 22, 2023
Merged

[pull] master from microsoft:master#58
pull[bot] merged 3 commits intospecialized806:masterfrom
deepspeedai:master

Conversation

@pull
Copy link
Copy Markdown

@pull pull bot commented Nov 22, 2023

See Commits and Changes for more details.


Created by pull[bot]

Can you help keep this open source service alive? 💖 Please sponsor : )

PareesaMS and others added 3 commits November 17, 2023 12:06
Co-authored-by: Ammar Ahmad Awan <ammar.awan@microsoft.com>
In case stage3 prompts are too long, the prompts are used but they are
arbitrary sliced at start to fit into the configured max prompt length.
This arbitrary slicing sometimes causes prompts to be less meaningful.
Which in turn, causes the generator to generate garbage.
This phenomena was observed to de-stabilize RLHF stage3.
To fix it, we filter prompts that are too long.

In addition, dataset rebuild flag is propagated to other required consumers.
Note that since generated dataset are cached in disk, this commit will have
effect only if we cleanup step3 cached datasets.

Change-Id: I440f09decf0784e4c2c8167a893006dff312281b

Signed-off-by: Moshe Island <misland@habana.ai>
Co-authored-by: Moshe Island <misland@habana.ai>
@pull pull bot added the ⤵️ pull label Nov 22, 2023
@pull pull bot merged commit b116838 into specialized806:master Nov 22, 2023
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants