Closed sfahad1414 closed 2 weeks ago
The changes in this pull request encompass modifications across multiple files, including reordering import statements and adjusting logging configurations in kscheduler.py
, introducing new logic in the constructor of the class in processor.py
, adding a new dependency in requirements/prod.txt
, expanding test coverage in action_service_test.py
, and updating the YAML schema for AI model configurations in llm_metadata.yml
. These updates collectively enhance code organization, flexibility, and testing capabilities without altering existing functionalities.
Files | Change Summary |
---|---|
kairon/events/scheduler/kscheduler.py | Reordered import statements for clarity; adjusted logging configuration to set 'apscheduler' to DEBUG and established basic logging setup. |
kairon/shared/llm/processor.py | Introduced new logic in the constructor for llm_secret_embedding based on llm_type ; updated get_embedding method accordingly. Added import for DEFAULT_LLM . |
requirements/prod.txt | Added new dependency google-generativeai . |
tests/integration_test/action_service_test.py | Added new test function to validate prompt action response with different embedding and completion strategies. |
tests/integration_test/services_test.py | Updated assertions in tests to replace "claude" with "anthropic" for LLM metadata validation. |
tests/testing_data/llm_metadata.yml | Renamed section from claude to anthropic ; modified max_tokens property and updated model enumeration values. |
tests/unit_test/llm_test.py | Changed the argument in LLMProcessor constructor for clarity in test case naming. |
kairon/shared/llm/processor.py
in both the main PR and this retrieved PR involve modifications to the handling of LLM metadata, indicating a direct connection in the code related to LLM processing and metadata management.In the code where bunnies hop,
Imports dance and never stop.
With logs that sing in DEBUG light,
Our scheduler's now a joyful sight!
So here’s to changes, neat and bright,
A happy code, a coder's delight! 🐇✨
[!TIP]
Announcements
- The review status is no longer posted as a separate comment when there are no actionable or nitpick comments. In such cases, the review status is included in the walkthrough comment. - We have updated our review workflow to use the Anthropic's Claude family of models. Please share any feedback in the [discussion post](https://discordapp.com/channels/1134356397673414807/1279579842131787838) on our Discord. - Possibly related PRs: Walkthrough comment now includes a list of potentially related PRs to help you recall past context. Please share any feedback in the [discussion post](https://discordapp.com/channels/1134356397673414807/1282535539299323995) on our Discord. - Suggested labels: CodeRabbit can now suggest labels by learning from your past PRs in the walkthrough comment. You can also provide custom labeling instructions in the UI or configuration file. - Possibly related PRs, automatic label suggestions based on past PRs, learnings, and possibly related issues require data opt-in (enabled by default).
tests/integration_test/services_test.py (3)
`1235-1237`: **LGTM!** The code changes are approved. The assertions correctly check for the presence of "anthropic" in the `actual["data"]` dictionary and verify the properties associated with it. --- `1252-1259`: **LGTM!** The code changes are approved. The test data is correctly updated by replacing references to "claude" with "anthropic". --- `1284-1286`: **LGTM!** The code changes are approved. The assertions correctly check for the presence of "anthropic" in the `actual["data"]` dictionary and verify the properties associated with it.
Summary by CodeRabbit