Ветки

Merge pull request #267 from whylabs/validation-update Added recommended preset validation for refusals

Update version to 0.0.31 (#248) Co-authored-by: w0-automator <whylabs-automator@users.noreply.github.com>

Add init method to token metric

mypy

branch to evaluate injection vector db

Add pandas-stubs types and some usage examples in langkit/metrics/__init__.py Noticed that the types in vscode were different than vim and then learned that vscode packages the pandas-stubs package by default, even if you don't use it. I added the last version of this package that supports python 3.8 here. The types aren't great but if it's what everyone using vscode will be seeing by default then I think its worth respecting.

action must exist on mainline before we can experiment with it on a branch

supported llm notes in hallucination and proactive injection detection

Add support for vaderSentiment

azure llm endpoint draft

change nb to markdown to better visualize/edit

specify schema name to register UDFs into

Rename language configuration example notebook and update pip install command

Another multilingual approach

Update README.md

experimental multilingual idea

Update injection examples and tests (#170) * Update injection examples and tests * Support either name in the notebook example of injection metric

Add abstraction aroun openai invocation params and make llm integration configurable

Use pytest.approx for asserting similarity scores to avoid fintermittent failing

crlf

fix patterns test

Sync version info to latest release on repo 0.0.11

version bump to 0.0.11

Update version to 0.0.9

add nlp evaluation score metrics

reorder emoji test

another swing

Delete sentiment and toxicity notebook

Delete logging text notebook

Delete LLM to WhyLabs notebook