[lora] Add load option to LoRA adapter API #2536
Merged
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
xyang16
force-pushed
the
lora
branch
12 times, most recently
from
52d9886 to
7a3e234
Compare
siddvenk
reviewed
Nov 11, 2024
| if adapter_load: | ||
| _service.add_lora(adapter_name, adapter_alias, adapter_path) | ||
| else: | ||
| _service.remove_lora(adapter_name, adapter_alias) |
There was a problem hiding this comment.
if adapter_load is false, why are we removing the adapter? Should this just be a noop?
There was a problem hiding this comment.
This is to have an unload option.
There was a problem hiding this comment.
I see, so we support unloading only for unpinned adapters
siddvenk
reviewed
Nov 11, 2024
Comment on lines
286
to
287
| return self.engine.add_lora(lora_request) and self.engine.pin_lora( | ||
| lora_request.lora_int_id) |
There was a problem hiding this comment.
do we need to check the result of add_lora before pinning?
Also, I would prefer if we kept these calls separate. It's more readable.
Same questions for vlm_rolling_batch.py
There was a problem hiding this comment.
- This is to make sure it's successfully loaded before pinning.
- Made these calls separate.
siddvenk
approved these changes
Nov 12, 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Outdated suggestions cannot be applied.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.
Description
Add an additional
loadoption to the register adapter API and update adapter API.The reason is for this change is to keep consistent with other model servers like vllm and lorax. vllm load_lora_adapter API don't load adapter weights, the adapter weights is only loaded when running inference using one particular adapter.
Discussed with Hosting team, making default to
true.Example: