From the model card
Fields below are copied from the tags and counters on the HuggingFace repository mistralai/Mistral-7B-Instruct-v0.2 at our last fetch. They are set by the uploader, not verified by us; rows with no tag are omitted. How this page is made.
- Publisher (HF namespace)
- mistralai
- Pipeline tag
- text-generation
- Library
- Transformers
- Framework tags
- PyTorch
- Weight formats
- safetensors
- License tag
apache-2.0— read the license file in the repo before relying on it- Papers cited
- arXiv:2310.06825
- Downloads (HF counter at last fetch)
- 1,162,867
- Likes (HF counter at last fetch)
- 3,206
- Model card
- https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2
Use cases
- General-purpose chat and instruction following at 7B scale
- RAG applications needing 32K context with fast inference
- Fine-tuning base for instruction-following downstream tasks
- Local deployment on consumer hardware via GGUF
Pros
- 32K context window via sliding window attention
- Apache-2.0 licensed for commercial use
- Strong performance on instruction benchmarks relative to its size
- Excellent GGUF and AWQ quantization ecosystem
Cons
- Mistral-7B-v0.3 and Mistral-Nemo supersede it for most tasks
- Instruction following can be inconsistent on complex multi-constraint prompts
- No native tool/function calling without further fine-tuning
- Lags Llama 3.1 8B Instruct on coding and math benchmarks
Tags
transformerspytorchsafetensorsmistraltext-generationfinetunedmistral-commonconversationalarxiv:2310.06825license:apache-2.0eval-resultstext-generation-inferenceregion:usdeploy:azure