BUG: Correct the input bytes data by langchain_openai #2589

[xiyuan-lee]

No description provided.

[qinxuye]

Lint failed, please install pre-commit and run pre-commit install at the project directory, this will do pre-check before committing.

[xiyuan-lee]

what's wrong?

2024-11-28 16:40:53.802 [info] An error has occurred: InvalidManifestError:
==> File /home/oem/.cache/pre-commit/repoq8t72eld/.pre-commit-hooks.yaml
==> At Hook(id='black')
==> At key: stages
==> At index 0
=====> Expected one of commit, commit-msg, manual, merge-commit, post-checkout, post-commit, post-merge, post-rewrite, prepare-commit-msg, push but got: 'pre-commit'
Check the log at /home/oem/.cache/pre-commit/pre-commit.log

[qinxuye]

what's wrong?

2024-11-28 16:40:53.802 [info] An error has occurred: InvalidManifestError: ==> File /home/oem/.cache/pre-commit/repoq8t72eld/.pre-commit-hooks.yaml ==> At Hook(id='black') ==> At key: stages ==> At index 0 =====> Expected one of commit, commit-msg, manual, merge-commit, post-checkout, post-commit, post-merge, post-rewrite, prepare-commit-msg, push but got: 'pre-commit' Check the log at /home/oem/.cache/pre-commit/pre-commit.log

You need to cd the xinference directory, there is a config file of pre-commit.

[qinxuye]

I just found this part is implemented in rest API, maybe it's better to put them into the embedding model?

inference/xinference/model/embedding/core.py

Line 226 in 0d4cb9c

def create_embedding(self, sentences: Union[str, List[str]], **kwargs):

Since rest API may not be in charge of any computation including the decoding.

[xiyuan-lee]

I just found this part is implemented in rest API, maybe it's better to put them into the embedding model?

inference/xinference/model/embedding/core.py

Line 226 in 0d4cb9c

def create_embedding(self, sentences: Union[str, List[str]], **kwargs):

Since rest API may not be in charge of any computation including the decoding.

Ok, I move the changes to the embedding model.

[qinxuye]

I slightly modified your code, check all the data if all of them is int is not so necessary IMHO, I think we could only check if it's list of list of int, and check the first int is OK, or this check may cost too much CPU.

[qinxuye]

LGTM