Fix excessive CPU memory consumption on TGI startup #595
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
dacorvo
force-pushed
the
launch_tgi_xlarge
branch
from
ac88ffb
to
fe71226
Compare
michaelbenayoun
approved these changes
May 16, 2024
| @@ -91,7 +91,7 @@ def fetch_model( | |||
| # Prefetch the neuron model from the Hub | |||
| logger.info(f"Fetching revision [{revision}] for neuron model {model_id} under {HF_HUB_CACHE}") | |||
| log_cache_size() | |||
| return snapshot_download(model_id, revision=revision) | |||
| return snapshot_download(model_id, revision=revision, ignore_patterns="*.bin") | |||
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Outdated suggestions cannot be applied.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.
What does this PR do?
When launching a TGI instance with a non-neuron model as parameter, the model needs to be exported from cached neuron artifacts during the container startup.
Before this change, the export was done without minimizing the CPU memory, which made it impossible to use this kind of "on-the-fly" export on the smaller
ml.inf2.xlargeinstances.