LLM EXPLORER
 LLM Explorer: A Curated Large Language Model Directory and Analytics
Dark Theme
  LLM  List    LLM Hosting    LLM Leaderboards    Blog    Newsfeed    Advertise

Mistrilitary 7B by Heralax

 »  All LLMs  »  Heralax  »  Mistrilitary 7B   URL Share it on

  128   
  Autotrain compatible Base model:heralax/army-pretra... Base model:quantized:heralax/a...   Conversational   Endpoints compatible   Generated from trainer   Ggml   Gguf   Mistral   Pytorch   Q8   Quantized   Region:us
Model Card on HF 🤗: https://huggingface.co/Heralax/Mistrilitary-7b 

Mistrilitary 7B Benchmarks

ARC: 66.55 vs 96.7 (so35)-31.2%
HellaSwag: 85.82 vs 95.3 (gpt4)-9.9%
MMLU: 64.63 vs 88.3 (so35)-26.8%
TruthfulQA: 56.8 vs 59 (gpt4)-3.7%
WinoGrande: 79.32 vs 87.5 (gpt4)-9.3%
GSM8K: 66.72 vs 96.4 (so35)-30.8%
LLME Score: 0.24022
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Mistrilitary 7B (Heralax/Mistrilitary-7b)

Mistrilitary 7B Parameters and Internals

Use Cases 
Primary Use Cases:
Question Answering based on US Army field manuals
Limitations:
May be inflexible to non-QA tasks
Additional Notes 
Data was mostly generated by a smaller model, Mistral NeMo, for cost efficiency. No generalist assistant data used. Experimental focus on army field manuals due to the abundance of such texts.
Training Details 
Data Sources:
US Army field manuals
Methodology:
Domain-expert finetune focused on factual QA
Hardware Used:
5 GPUs
Input Output 
Performance Tips:
Low temperature recommended, screenshots use 0
LLM NameMistrilitary 7B
Repository 🤗https://huggingface.co/Heralax/Mistrilitary-7b 
Base Model(s)  Heralax/army-pretrain-1   Heralax/army-pretrain-1
Model Size7b
Required VRAM7.7 GB
Updated2025-02-22
MaintainerHeralax
Model Typemistral
Model Files  14.5 GB   7.7 GB   14.5 GB
GGML QuantizationYes
GGUF QuantizationYes
Quantization Typeggml|q8|gguf
Model ArchitectureMistralForCausalLM
Licenseapache-2.0
Context Length32768
Model Max Length32768
Transformers Version4.45.0
Tokenizer ClassLlamaTokenizer
Padding Token<|end_of_text|>
Vocabulary Size32001
Torch Data Typebfloat16

Best Alternatives to Mistrilitary 7B

Best Alternatives
Context / RAM
Downloads
Likes
Mistral Scary Story Finetune32K / 14.5 GB300
Mistral 7B V0.2 Csn SFT32K / 14.4 GB150
AzzurroQuantized32K / 4.4 GB3334
Instruct 16bit32K / 0.4 GB50
Merlinite 7B Ocp4.15 V0.332K / 14.5 GB90
Shiftdocs 7B Ocp4.15 V0.332K / 14.5 GB110
AzzurroQuantized32K / 4.4 GB2224
Merlinite 7B Ocp4.15 V0.132K / 4.4 GB120
Filiberto 7B Instruct Exp132K / 14.5 GB310
MistralMusic32K / 14.4 GB261

Rank the Mistrilitary 7B Capabilities

🆘 Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! 🌟

Instruction Following and Task Automation  
Factuality and Completeness of Knowledge  
Censorship and Alignment  
Data Analysis and Insight Generation  
Text Generation  
Text Summarization and Feature Extraction  
Code Generation  
Multi-Language Support and Translation  

What open-source LLMs or SLMs are you in search of? 43470 in total.

Email us: info@extractum.io. Our Privacy Policy  | Terms and Conditions  | Suggest an improvement.
Our Social Media →  
Original data from HuggingFace, OpenCompass and various public git repos.
Release v20241227
  Support LLM Explorer