search

opensearch-project / ml-commons

ml-commons provides a set of common machine learning algorithms, e.g. k-means, or linear regression, to help developers build ML related features within OpenSearch.
Apache License 2.0
98 stars 136 forks source link

[FEATURE] Support for Code Interpreter Tool like E2B so that we can run any Python Code using Interpreter tool once code generated by LLM #2493

Closed ramda1234786 closed 5 months ago

ramda1234786 commented 5 months ago

Is your feature request related to a problem? With Agents not able to run code generated by LLM. Just like we have Tools for Index, Mapping, VectorDB, RAG, we can also have it for Code Interpreter Tool.

One example is here https://github.com/e2b-dev/code-interpreter

What solution would you like? Once the LLM Generated the Code, we should be run them. To start with Python Language Should be able to Run the Code also when passing the code to the Agentic Workflow

What alternatives have you considered? Using some other Agentic Tool like LlamaIndex with OpenSearch Vector Index, but this does not sound to be good solution

Do you have any additional context? With this code interpreter we should be able to check the code response live by running them with the Agent Itself

ylwu-amzn commented 5 months ago

We are building a connector tool which allows Agent to call external service(https://github.com/opensearch-project/ml-commons/issues/2497). If this Code Interpreter has REST API (or you can simply wrap it into a REST API), then you can use the connector tool to call external service.

ramda1234786 commented 5 months ago

okay great. Thanks

ylwu-amzn commented 5 months ago

@ramda1234786 I will close this one as we are going to build Connector Tool (https://github.com/opensearch-project/ml-commons/issues/2497) which can support this. Feel free to open if you think it's going to be a different feature.