---
license: apache-2.0
datasets:
- abacusai/MetaMathFewshot
- shahules786/orca-chat
- anon8231489123/ShareGPT_Vicuna_unfiltered
---

Trained on the MetamathFewshot (https://huggingface.co/datasets/abacusai/MetaMathFewshot) dataset from base Mistral, as well as the Vicuna (https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered) dataset and the OrcaChat (https://huggingface.co/datasets/shahules786/orca-chat) dataset.

Instruction tuned with the following parameters:

- LORA, Rank 8, Alpha 16, Dropout 0.05, all modules (QKV and MLP)
- 3 epochs
- Micro Batch Size 32 over 4xH100, gradient accumulation steps = 1
- AdamW with learning rate 5e-5