DIBT-Mistral-7B-QLoRA
Mistral 7B QLoRA fine-tuned on my dibt-instruct dataset.
Model Details
Model Description
Mistral 7B fine-tuned on my dibt-instruct dataset. This model has a unique speaking style compared to most Mistral fine-tunes. Not sure if people prefers it but it's certainly different.
- Developed by: Locutusque
- Funded by : Kaggle, Google
- Shared by : Hugging Face
- Model type: Decoder-only transformer
- Language(s) (NLP): English
- License: apache-2.0
Uses
This model is one of the experts used in the MoE model "Hyperion-3.0-Mixtral-3x7B"
Direct Use
General-purpose question answering. Can do a bit of coding.
Bias, Risks, and Limitations
This model struggles sometimes with math, unless prompted with chain of thought. Inherets the writing style of Gemini Pro.
- Downloads last month
- 0
Inference Providers
NEW
This model isn't deployed by any Inference Provider.
🙋
Ask for provider support