Skip to content
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.

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

Fix Gemma 7B checkpoint save #1169

Merged
merged 1 commit into from
Jul 12, 2024

Conversation

ebsmothers
Copy link
Contributor

@ebsmothers ebsmothers commented Jul 12, 2024

Fixes #1122. See point (2) here.

Test:

tune run --nnodes 1 --nproc_per_node 4 lora_finetune_distributed --config \
gemma/7B_lora max_steps_per_epoch=5 epochs=1
...
INFO:torchtune.utils.logging:Model checkpoint of size 5.00 GB saved to /tmp/gemma-7b/hf_model_0001_0.pt
INFO:torchtune.utils.logging:Model checkpoint of size 4.98 GB saved to /tmp/gemma-7b/hf_model_0002_0.pt
INFO:torchtune.utils.logging:Model checkpoint of size 4.98 GB saved to /tmp/gemma-7b/hf_model_0003_0.pt
INFO:torchtune.utils.logging:Model checkpoint of size 2.11 GB saved to /tmp/gemma-7b/hf_model_0004_0.pt
INFO:torchtune.utils.logging:Adapter checkpoint of size 0.37 GB saved to /tmp/gemma-7b/adapter_0.pt
INFO:torchtune.utils.logging:Adapter checkpoint of size 0.37 GB saved to /tmp/gemma-7b/adapter_model.bin
INFO:torchtune.utils.logging:Adapter checkpoint of size 0.00 GB saved to /tmp/gemma-7b/adapter_config.json

Copy link

pytorch-bot bot commented Jul 12, 2024

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/torchtune/1169

Note: Links to docs will display an error until the docs builds have been completed.

✅ No Failures

As of commit 7033610 with merge base 1af5135 (image):
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@facebook-github-bot facebook-github-bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Jul 12, 2024
Copy link
Contributor

@joecummings joecummings left a comment

Choose a reason for hiding this comment

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

show me the receipts

@joecummings
Copy link
Contributor

show me the receipts

(Screenshot of successful model + adapter saving)

@joecummings
Copy link
Contributor

show me the receipts

(Screenshot of successful model + adapter saving)

Wasn't sure if I was being clear

@joecummings
Copy link
Contributor

show me the receipts

(Screenshot of successful model + adapter saving)

Wasn't sure if I was being clear

Does it make sense?

@ebsmothers
Copy link
Contributor Author

show me the receipts

(Screenshot of successful model + adapter saving)

Wasn't sure if I was being clear

Does it make sense?

No

@ebsmothers
Copy link
Contributor Author

show me the receipts

(Screenshot of successful model + adapter saving)

Wasn't sure if I was being clear

Does it make sense?

No

But I updated the test plan

@joecummings joecummings merged commit cc92fa0 into pytorch:main Jul 12, 2024
29 checks passed
maximegmd pushed a commit to maximegmd/torchtune that referenced this pull request Jul 13, 2024
pbontrager pushed a commit that referenced this pull request Jul 15, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Missing lm_head.weight key when using Gemma 7B distributed LoRA recipe with gemma-7b-it
3 participants