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

Zephyr Python Ru by MexIvanov

 »  All LLMs  »  MexIvanov  »  Zephyr Python Ru   URL Share it on

  0   

Zephyr Python Ru is an open-source language model by MexIvanov. Features: 7b LLM, VRAM: 0.1GB, License: mit, HF Score: 60.1, LLM Explorer Score: 0.13, Arc: 56.1, HellaSwag: 82, MMLU: 60.2, TruthfulQA: 52.8, WinoGrande: 76.8, GSM8K: 32.5.

  Arxiv:2409.09353   Adapter Base model:finetune:huggingfac... Base model:huggingfaceh4/zephy...   Conversational Dataset:mexivanov/codeexercise... Dataset:mexivanov/vezora-teste... Dataset:zelkame/ru-stackoverfl...   En   Finetuned   Lora   Region:us   Ru   Safetensors
Model Card on HF 🤗: https://huggingface.co/MexIvanov/zephyr-python-ru 

Zephyr Python Ru Benchmarks

ARC: 56.14 vs 96.7 (so35)-41.9%
HellaSwag: 82.03 vs 95.3 (gpt4)-13.9%
MMLU: 60.18 vs 88.3 (so35)-31.8%
TruthfulQA: 52.8 vs 59 (gpt4)-10.5%
WinoGrande: 76.8 vs 87.5 (gpt4)-12.2%
GSM8K: 32.52 vs 96.4 (so35)-66.3%
LLME Score: 0.12926
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Zephyr Python Ru (MexIvanov/zephyr-python-ru)
🌟 Advertise your project 🚀

Zephyr Python Ru Parameters and Internals

Model Type 
LoRA adapter, Peft
Use Cases 
Primary Use Cases:
Instruction-based coding in Python, based on instructions written in English or Russian
Limitations:
Model is not aligned to human preferences for safety, No moderation mechanisms, Trained on code based instruction so may produce problematic outputs without filtering
Considerations:
Users should be aware of risks, biases, and limitations of the model.
Additional Notes 
This adapter model was trained using `bitsandbytes` quantization config: 4-bit load, nf4 quant type, float16 compute dtype.
Supported Languages 
ru (high), en (high), Python (high)
Training Details 
Data Sources:
zelkame/ru-stackoverflow-py, MexIvanov/Vezora-Tested-22k-Python-Alpaca-ru, MexIvanov/CodeExercise-Python-27k-ru
Methodology:
Special training methods or approaches used, such as fine-tuning techniques.
Model Architecture:
A LoRA (Peft) adapter model trained on a mix of publicly available data and machine-translated synthetic python coding datasets.
Responsible Ai Considerations 
Mitigation Strategies:
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model.
Input Output 
Input Format:
<|system|> ~~ <|user|> {prompt}~~ <|assistant|>
Accepted Modalities:
text
LLM NameZephyr Python Ru
Repository 🤗https://huggingface.co/MexIvanov/zephyr-python-ru 
Base Model(s)  Zephyr 7B Beta   HuggingFaceH4/zephyr-7b-beta
Model Size7b
Required VRAM0.1 GB
Updated2026-04-03
MaintainerMexIvanov
Model Files  0.1 GB   0.1 GB
Supported Languagesen ru
Model ArchitectureAdapter
Licensemit
Model Max Length1024
Is Biasednone
Tokenizer ClassLlamaTokenizer
Padding Token</s>
PEFT TypeLORA
LoRA ModelYes
PEFT Target Modulesv_proj|q_proj
LoRA Alpha16
LoRA Dropout0.05
R Param64

Best Alternatives to Zephyr Python Ru

Best Alternatives
Context / RAM
Downloads
Likes
Qwen Megumin0K / 0.1 GB51
...s 25 Mistral 7B Irca DPO Pairs0K / 0.1 GB50
Uk Fraud Chatbot Llama20K / 0.4 GB50
Qwen1.5 7B Chat Sa V0.10K / 0 GB50
Zephyr 7B Ipo 0K 15K I10K / 0.7 GB460
Deepseek Llm 7B Chat Sa V0.10K / 0 GB50
Hr Other 7B Lora0K / 0.2 GB300
Deepthink Reasoning Adapter0K / 0.2 GB33
Mistral 7B Instruct Sa V0.10K / 0 GB90
... Days Of Sodom LoRA Mistral 7B0K / 0.2 GB50
Note: green Score (e.g. "73.2") means that the model is better than MexIvanov/zephyr-python-ru.

Rank the Zephyr Python Ru 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? 52721 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.
Check out Ag3ntum — our secure, self-hosted AI agent for server management.
Release v20260328a
  Support LLM Explorer