Skip to content

[WIP] Add state_dict converter for DeepSeekv3 in torchtitan #1538

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Draft
wants to merge 6 commits into
base: main
Choose a base branch
from
Draft

[WIP] Add state_dict converter for DeepSeekv3 in torchtitan #1538

wants to merge 6 commits into from

Conversation

wwwjn
Copy link
Contributor

@wwwjn wwwjn commented Aug 6, 2025 •
edited
Loading

Support loading a DeepSeek HF weights to Deepseek-V3 model:

  1. Support split / concat weight for GroupedExperts
  2. Support _dequantization during to_hf()

@meta-cla meta-cla bot added the CLA Signed This label is managed by the Meta Open Source bot. label Aug 6, 2025
wwwjn
wwwjn commented Aug 7, 2025
View reviewed changes
torchtitan/models/deepseek_v3/model/quantization.py
col_end = min(col_start + BLOCK_SIZE, orig_shape[1])

# Get the block
block = float_weight[row_start:row_end, col_start:col_end]
Copy link
Contributor Author

@wwwjn wwwjn Aug 7, 2025 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Here's the tensor slice operation @ankitageorge

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@tianyu-l tianyu-l Awaiting requested review from tianyu-l tianyu-l will be requested when the pull request is marked ready for review tianyu-l is a code owner

@fegin fegin Awaiting requested review from fegin fegin will be requested when the pull request is marked ready for review fegin is a code owner

@wconstab wconstab Awaiting requested review from wconstab wconstab will be requested when the pull request is marked ready for review wconstab is a code owner

At least 1 approving review is required to merge this pull request.

Assignees
No one assigned
Labels
CLA Signed This label is managed by the Meta Open Source bot.
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
1 participant
@wwwjn