gorilla
gorilla copied to clipboard
Gorilla: Training and Evaluating LLMs for Function Calls (Tool Calls)
[ActionGemma](https://huggingface.co/KishoreK/ActionGemma-9B) is a Gemma2 9b based fine-tuned model for function calling. This PR adds model handler for the it.
Add huggingface handler for BFCL. Test: TBD
**Describe the issue** There are several format issues in the possible answers of live test cases. **ID datapoint** 1. Datapoint / Model Handler permalink: https://github.com/ShishirPatil/gorilla/blob/main/berkeley-function-call-leaderboard/data/possible_answer/BFCL_v2_live_parallel_multiple.json Here are failed examples from...
**Description** The instructions in Java/Javascript test cases often request explanations or descriptions of function calls rather than directly requesting the construction of function calls. This leads to failures in the...
The Llama31 series models have all been released. May I ask if you have plans to test the effectiveness of the Llama31 8B, 70B, and 405B models?
[Llama-3.1-Storm-8B](https://huggingface.co/blog/akjindal53244/llama31-storm8b) model was recently released. This PR adds model handler for Llama-3.1-Storm-8B.
In the `get_chunks` function, the PDF pages are not separated, yielding chunks where spaces, dots or new lines are missing. https://github.com/ShishirPatil/gorilla/blob/2fc82a9bd1b667b594504556a58d0e8ee908c592/raft/raft.py#L90-L92 Unclear how to properly handle this situation in a...
**Is the feature request related to a problem?** No, OpenAI recently released new feature to support structured output with constrained decoding, would love to see the standing on the leaderboard...