library_name: transformers | |
inference: | |
parameters: | |
temperature: 1 | |
top_p: 0.95 | |
top_k: 40 | |
repetition_penalty: 1.2 | |
license: apache-2.0 | |
language: | |
- en | |
pipeline_tag: text-generation | |
[](https://hf.co/QuantFactory) | |
# QuantFactory/Ministral-3b-instruct-GGUF | |
This is quantized version of [ministral/Ministral-3b-instruct](https://huggingface.co/ministral/Ministral-3b-instruct) created using llama.cpp | |
# Original Model Card | |
 | |
### Model Description | |
<!-- Provide a longer summary of what this model is. --> | |
Ministral is a series of language model, build with same architecture as the famous Mistral model, but with less size. | |
- **Model type:** A 3B parameter GPT-like model fine-tuned on a mix of publicly available, synthetic datasets. | |
- **Language(s) (NLP):** Primarily English | |
- **License:** Apache 2.0 | |
- **Finetuned from model:** [mistralai/Mistral-7B-v0.1](https://huggingface.co/mistralai/Mistral-7B-v0.1) | |