LLM EXPLORER 65,548 MODELS INDEXED

BioMistral 7B DARE by BioMistral

By BioMistral · 1354 downloads
HF Repo ↗ Copy URL

BioMistral 7B DARE is an open-source language model by BioMistral. Features: 7b LLM, VRAM: 14.4GB, Context: 32K, License: apache-2.0, Instruction-Based, Merged, LLM Explorer Score: 0.14, Arc: 58.3, HellaSwag: 79.9, MMLU: 57.3, GSM8K: 15.

  Merged Model   Arxiv:2306.01708   Arxiv:2311.03099   Arxiv:2402.10373 Base model:biomistral/biomistr... Base model:mistralai/mistral-7...   Biology   Conversational   Dare   Dataset:pubmed   De   Deploy:azure   En   Endpoints compatible   Es   Fr   Instruct   It   Medical   Mistral   Nl   Pl   Region:us   Ro   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/BioMistral/BioMistral-7B-DARE 

BioMistral 7B DARE Parameters and Internals

Model Type 
text-generation
Use Cases 
Areas:
research
Applications:
biomedical research, text generation
Primary Use Cases:
medical question-answering, multilingual evaluation
Limitations:
Not suitable for professional medical advice, Potential biases and ethical issues
Considerations:
Users need to be aware of biases and test extensively before use in critical scenarios.
Additional Notes 
The model and its evaluation benchmarks are freely available for research purposes.
Supported Languages 
en (English), fr (French), nl (Dutch), es (Spanish), it (Italian), pl (Polish), ro (Romanian), de (German)
Training Details 
Data Sources:
PubMed Central
Methodology:
DARE (using TIES merge method) on base model Mistral-7B-Instruct-v0.1
Context Length:
2048
Hardware Used:
CNRS Jean Zay French HPC
Model Architecture:
Mistral-based architecture
Responsible Ai Considerations 
Fairness:
Under evaluation
Transparency:
Requires user caution due to potential biases
Accountability:
Developers emphasize research use over production use
Mitigation Strategies:
Advised testing in random controlled trials for medical applications
Input Output 
Input Format:
Text
Accepted Modalities:
text
Output Format:
Text
LLM NameBioMistral 7B DARE
Repository 🤗https://huggingface.co/BioMistral/BioMistral-7B-DARE 
Base Model(s)  BioMistral 7B   mistralai/Mistral-7B-Instruct-v0.1   BioMistral/BioMistral-7B   mistralai/Mistral-7B-Instruct-v0.1
Merged ModelYes
Model Size7b
Required VRAM14.4 GB
Updated2026-10-21
MaintainerBioMistral
Model Typemistral
Instruction-BasedYes
Model Files  9.9 GB: 1-of-2   4.5 GB: 2-of-2
Supported Languagesen fr nl es it pl ro de
Model ArchitectureMistralForCausalLM
Licenseapache-2.0
Context Length32768
Model Max Length32768
Transformers Version4.36.2
Tokenizer ClassLlamaTokenizer
Vocabulary Size32000
Torch Data Typebfloat16

Quantized Models of the BioMistral 7B DARE

Model
Likes
Downloads
VRAM
BioMistral 7B DARE GGUF23172 GB

Best Alternatives to BioMistral 7B DARE

Best Alternatives
Context / RAM
Downloads
Likes
...Nemo Instruct 2407 Abliterated1000K / 24.5 GB22020
SpydazWeb AI HumanAI RP512K / 14.4 GB161
SpydazWeb AI HumanAI 002512K / 14.4 GB181
...daz Web AI ChatML 512K Project512K / 14.5 GB120
... Summarize 64K QLoRANET Merged128K / 4.1 GB140
...1 Summarize 64K LoRANET Merged128K / 14.4 GB230
Mistral 7B Instruct V0.232K / 14.4 GB11153373184
Mistral 7B Instruct V0.132K / 14.4 GB2536291846
Sentiment Analysis 7B32K / 14.5 GB66122
Mistral 7B Instruct V0.232K / 14.4 GB13380
Note: green Score (e.g. "73.2") means that the model is better than BioMistral/BioMistral-7B-DARE.