Add web endpoint to the inference server #38
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
3658c66
to
3aa7a9e
Compare
|
I think the run_name change makes sense even just on its own but can we update the readme too? |
mwaskom
reviewed
Mar 12, 2024
src/inference.py
Outdated
| config = yaml.safe_load(f.read()) | ||
| model_path = (Path(run_folder) / config["output_dir"] / "merged").resolve() | ||
|
|
||
| def __init__(self, run_name: str = "", model_dir: str = "/runs") -> None: |
There was a problem hiding this comment.
model_dir is a little confusing since the two things we store in volumes are (1) pretrained models and (2) finetuned models. They’re all models :)
There was a problem hiding this comment.
but everything under /runs is finetuned models right?
There was a problem hiding this comment.
Arguably not since there’s config, training logs, metrics, etc. but it’s a pretty nitpicky point either way
There was a problem hiding this comment.
ok i'll rename it to run_dir
mwaskom
reviewed
Mar 12, 2024
src/inference.py
Outdated
| run_name = VOLUME_CONFIG[self.model_dir].listdir("/")[-1].path | ||
|
|
||
| # Grab the output dir (usually "lora-out") | ||
| with open(f"/runs/{run_name}/config.yml") as f: |
There was a problem hiding this comment.
Looks like we have /runs hardcoded here even though it’s a parameter?
3aa7a9e
to
29ca3b7
Compare
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Outdated suggestions cannot be applied.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.
This lets you do
modal serve src.inferencein one window, and
curl 'https://modal-labs--example-axolotl-inference-web-dev.modal.run?input=what+time+is+it'in another windowit streams!
Note that most of the complexity here is that we want to support an unparametrized class constructor, so we need to pick the run id automatically. In order to do that it was slightly easier to change the argument to be
run_namerather than the full path. Ifrun_nameisn't provided (which it never is in the case of web endpoints, since we don't support parametrized functions with web endpoints), then it just picks the last run by doing a directory listing.