[RFC] Configuring low precision training in torchtune #504
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/torchtune/504
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 New FailureAs of commit 5a037e9 with merge base 9c75d48 ( NEW FAILURE - The following job has failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
facebook-github-bot
added
the
CLA Signed
✅ Deploy Preview for torchtune-preview ready!
To edit notification comments on pull requests, go to your Netlify site configuration. |
rohan-varma
commented
Mar 15, 2024
|
|
||
| #### TL;DR | ||
| - Single device (both full finetune and LoRA) and multi device will support a flag, `dtype`, that can be either [bf16, fp32] to configure low precision training. This will initialize the model in the lower precision, so all parameters, activations, gradients, and opt. states will be in this precision to optimize memory savings. We will not enable torch.autocast. | ||
| - We will actively de-invest in fp16 training since most recent HW has support for bf16. In terms of consumer HW arches we'd like to support, 4090, 3090, A6000 support bf16, only T4 does not support bf16. This means in particular we won't have a memory efficient ( < 16GB) finetuning solution that runs reliably on T4's (but we will for the other mentioned consumer GPUs). |
There was a problem hiding this comment.
de-invest in fp16 for MVP specifically
rohan-varma
changed the title
This was referenced Mar 16, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Context
Changelog
Test plan