Yuki Watanabe

@MarkBusschers You can use [MlflowClient().log_figure()](https://mlflow.org/docs/latest/api_reference/python_api/mlflow.client.html#mlflow.client.MlflowClient.log_figure) to log an image with a specific Run. ``` client = mlflow.MlflowClient() client.log_figure(run_id=..., ) ``` Conventionally `mlflow.log_xyz` API (so called "fluent" API) focuses on logging...

@alexandre-daniel-ext Thank you for reporting the issue! MLflow does not support saving the raw agent runnable. Instead, it needs to be wrapped with `AgentExecutor` instance, as described in [the LangChain...

@alexandre-daniel-ext > The turnaround do not works because you can not make inference on an agent object (and indeed on an agent log in MLflow) : you must have an...

@Koenkk Thank you for raising the issue and proactively submit the fix! ~The change looks good to me🙂~ Sorry I overlooked one corner case., added a comment.

Aggregation of the metrics is supported in the recent version: https://mlflow.org/blog/deep-learning-part-1#metric-aggregation Kindly request updating mlflow and see if it satisfies the use case.

The proposal makes sense to me. @daniellok-db do you know if there is any reason for this upperbound?

@simonamtoft https://github.com/mlflow/mlflow/pull/13027 actually updated the numpy version to `

~~The condition issue is being addressed in https://github.com/mlflow/mlflow/pull/12592, with additional changes to handle new interface. Closing as duplicates.~~ Sorry the PR not change the upper bound itself. Reopening this one...

@ion-elgreco There are some conflict against main branch and format issues. Could you fix them so we can unblock this PR? * Please follow the PR template like [other PRs](https://github.com/mlflow/mlflow/pull/12996)....

@NevoleMarek Thank you for raising this issue. The proposal makes sense, we should make those compatible with OpenAI data models (with backward compatibility manner). Please feel free to submit a...