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

StableBeluga2 by stabilityai

 »  All LLMs  »  stabilityai  »  StableBeluga2   URL Share it on

  1902   
  Arxiv:2306.02707   Arxiv:2307.09288   Autotrain compatible Dataset:conceptofmind/cot subm... Dataset:conceptofmind/flan2021... Dataset:conceptofmind/niv2 sub... Dataset:conceptofmind/t0 submi...   En   Endpoints compatible   Llama   Pytorch   Region:us   Sharded
Model Card on HF 🤗: https://huggingface.co/stabilityai/StableBeluga2 

StableBeluga2 Benchmarks

MMLU Pro: 25.85
GPQA: 8.84
MUSR: 18.65
BBH: 41.26
IFEval: 37.87 vs 88 (so35)-57%
ARC: 91.2 vs 96.7 (so35)-5.7%
HellaSwag: 80.5 vs 95.3 (gpt4)-15.5%
MMLU: 68.6 vs 88.3 (so35)-22.3%
TruthfulQA: 59.44 vs 59 (gpt4)0.7%
WinoGrande: 82.95 vs 87.5 (gpt4)-5.2%
GSM8K: 69.6 vs 96.4 (so35)-27.8%
OpenCompas: 60.9 vs 73.3 (gpt4)-16.9%
Exam: 71.9 vs 77.2 (gpt4)-6.9%
Language: 44 vs 62 (gpt4)-29%
Knowledge: 62.2 vs 73.5 (gpt4)-15.4%
Understand: 72.5 vs 70 (gpt4)3.6%
Reasoning: 58 vs 74.4 (gpt4)-22%
MATH Lvl 5: 4.38
LLME Score: 0.22742
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
StableBeluga2 (stabilityai/StableBeluga2)
🌟 Advertise your project 🚀

StableBeluga2 Parameters and Internals

Model Type 
text generation, auto-regressive language model
Use Cases 
Areas:
research, commercial applications
Limitations:
Potential for producing inaccurate, biased, or objectionable responses.
Considerations:
Testing conducted in English, developers should perform additional safety testing before deployment.
Supported Languages 
English (full)
Training Details 
Data Sources:
conceptofmind/cot_submix_original, conceptofmind/flan2021_submix_original, conceptofmind/t0_submix_original, conceptofmind/niv2_submix_original
Methodology:
Supervised fine-tuning on an internal Orca-style dataset.
Model Architecture:
Fine-tuned Llama2 70B
Responsible Ai Considerations 
Fairness:
Beluga's potential outputs cannot be predicted in advance, and the model may produce inaccurate, biased or other objectionable responses.
Mitigation Strategies:
Developers should perform safety testing and tuning tailored to their specific applications.
Input Output 
Input Format:
System and User prompt
Accepted Modalities:
text
Output Format:
Assistant response
Performance Tips:
Use system prompts as recommended.
LLM NameStableBeluga2
Repository 🤗https://huggingface.co/stabilityai/StableBeluga2 
Required VRAM275.3 GB
Updated2025-08-20
Maintainerstabilityai
Model Typellama
Model Files  9.4 GB: 1-of-29   9.3 GB: 2-of-29   10.0 GB: 3-of-29   9.9 GB: 4-of-29   9.3 GB: 5-of-29   9.3 GB: 6-of-29   9.3 GB: 7-of-29   10.0 GB: 8-of-29   9.9 GB: 9-of-29   9.3 GB: 10-of-29   9.3 GB: 11-of-29   9.3 GB: 12-of-29   10.0 GB: 13-of-29   9.9 GB: 14-of-29   9.3 GB: 15-of-29   9.3 GB: 16-of-29   9.3 GB: 17-of-29   10.0 GB: 18-of-29   9.9 GB: 19-of-29   9.3 GB: 20-of-29   9.3 GB: 21-of-29   9.3 GB: 22-of-29   10.0 GB: 23-of-29   9.9 GB: 24-of-29   9.3 GB: 25-of-29   9.3 GB: 26-of-29   9.3 GB: 27-of-29   10.0 GB: 28-of-29   7.6 GB: 29-of-29
Supported Languagesen
Model ArchitectureLlamaForCausalLM
Context Length4096
Model Max Length4096
Transformers Version4.32.0.dev0
Tokenizer ClassLlamaTokenizer
Beginning of Sentence Token<s>
End of Sentence Token</s>
Unk Token<unk>
Vocabulary Size32000
Torch Data Typefloat32

Quantized Models of the StableBeluga2

Model
Likes
Downloads
VRAM
StableBeluga2 70B AWQ2736 GB
StableBeluga2 70B GGUF46129 GB
StableBeluga2 70B GPTQ902735 GB
StableBeluga2 70B GGML39428 GB

Best Alternatives to StableBeluga2

Best Alternatives
Context / RAM
Downloads
Likes
LWM Text 512K512K / 13.5 GB92
LWM Text Chat 512K512K / 13.5 GB72
LWM Text 256K256K / 13.5 GB83
LWM Text Chat 256K256K / 13.5 GB83
Pallas 0.5 LASER 0.1195K / 68.9 GB17642
Orpheus AUS128K /  GB110
Finetuning Health Ci128K / 6.5 GB52
Ukr Synth Phi 3.5128K / 7.6 GB50
WizardLM Phi 3.5128K / 7.6 GB50
Ashley3b X 1.2128K / 6.5 GB250
Note: green Score (e.g. "73.2") means that the model is better than stabilityai/StableBeluga2.

Rank the StableBeluga2 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? 50767 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 v20241124
  Support LLM Explorer