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

Atlas Chat 2B by MBZUAI-Paris

 »  All LLMs  »  MBZUAI-Paris  »  Atlas Chat 2B   URL Share it on

  632   
  Arxiv:2308.16149   Arxiv:2409.17912   Ar   Ary   Autotrain compatible Base model:finetune:google/gem... Base model:google/gemma-2-2b-i...   Conversational Dataset:mbzuai-paris/darija-sf...   Endpoints compatible   Gemma2   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/MBZUAI-Paris/Atlas-Chat-2B 

Atlas Chat 2B Benchmarks

LLME Score: 0.24087
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Atlas Chat 2B (MBZUAI-Paris/Atlas-Chat-2B)

Atlas Chat 2B Parameters and Internals

Model Type 
text generation
Use Cases 
Areas:
Research, Education, Commercial applications
Applications:
Conversational agents, Translation, Summarization, Text generation
Primary Use Cases:
Conversational interfaces, Text summarization, Research assistance
Limitations:
Biases in training data, Handling open-ended tasks
Considerations:
Potential socio-cultural biases and training data influences.
Additional Notes 
First open-source model to support Darija dialect.
Supported Languages 
ar (Darija (Moroccan Arabic))
Training Details 
Data Sources:
MBZUAI-Paris/Darija-SFT-Mixture
Data Volume:
450k instructions (max 2048 tokens)
Methodology:
Parameter-efficient fine-tuning with LoRA rank of 256
Context Length:
2048
Hardware Used:
8 Nvidia A100 80 GB GPUs
Model Architecture:
Atlas-Chat models based on Gemma 2
Responsible Ai Considerations 
Fairness:
LLMs trained on real-world text data can reflect socio-cultural biases.
Transparency:
Model card provides details on architecture, capabilities, and limitations.
Accountability:
Encouraged use of guidelines for responsible model use.
Mitigation Strategies:
Monitoring, evaluation metrics, human review, and de-biasing techniques encouraged.
Input Output 
Input Format:
Text prompts or questions
Accepted Modalities:
text
Output Format:
Generated Darija text
LLM NameAtlas Chat 2B
Repository 🤗https://huggingface.co/MBZUAI-Paris/Atlas-Chat-2B 
Base Model(s)  Gemma 2 2B It   google/gemma-2-2b-it
Model Size2b
Required VRAM5.2 GB
Updated2025-02-22
MaintainerMBZUAI-Paris
Model Typegemma2
Model Files  5.0 GB: 1-of-2   0.2 GB: 2-of-2
Supported Languagesar
Model ArchitectureGemma2ForCausalLM
Licensegemma
Context Length8192
Model Max Length8192
Transformers Version4.42.4
Tokenizer ClassGemmaTokenizer
Padding Token<pad>
Vocabulary Size256000
Torch Data Typebfloat16

Best Alternatives to Atlas Chat 2B

Best Alternatives
Context / RAM
Downloads
Likes
SJT 2B128K / 5.2 GB130
SILMA Kashif 2B Instruct V1.012K / 5.2 GB270414
Gemma 2 2B It8K / 5.2 GB421819908
Gemma 2 2B8K / 10.5 GB182986487
Gemma 2 2B Jpn It8K / 5.2 GB14320162
2 PRYMMAL ECE 2B SLERP V18K / 15.8 GB12290
...emma 2 2B It Chinese Kyara DPO8K / 15.7 GB763110
Gemma2Slerp2 2.6B8K / 5.3 GB1882
SJT 2B V1.18K / 5.3 GB501
GWQ2b8K / 5.2 GB6010
Note: green Score (e.g. "73.2") means that the model is better than MBZUAI-Paris/Atlas-Chat-2B.

Rank the Atlas Chat 2B 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