rinnaVicuna MRKL
Vicuna 13B LangChain MRKL delta
This research release fine-tunes Vicuna 13B to follow a strict LangChain MRKL format. It is distributed as a delta that must be applied to original LLaMA weights.
Research release
A formatting experiment with a narrow claim
The card describes 15 MRKL examples within mixed training data. Its authors state that they did not use an evaluation set and do not claim improved underlying reasoning ability.Vicuna 13B LangChain MRKL delta official model card
Distinguish action syntax from successful actions
Practical guidance. A correctly formatted tool request can still name the wrong tool or supply wrong arguments. Treat this release as a study of output formatting, and verify the surrounding action runner separately.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Status | Research release |
| Access | Downloadable |
| Model identifiers | vicuna-13b-delta-finetuned-langchain-MRKL |
| License declaration | The delta card declares CC BY-SA 4.0; the required original weights have separate terms.Vicuna 13B LangChain MRKL delta official model card |
| Ancestry | Vicuna 13B, derived from the original LLaMA family.Vicuna 13B LangChain MRKL delta official model card |