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

Qwen2 72B by Qwen

 »  All LLMs  »  Qwen  »  Qwen2 72B   URL Share it on

  11962   
  Autotrain compatible   Conversational   En   Endpoints compatible   Pretrained   Qwen2   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/Qwen/Qwen2-72B 

Qwen2 72B Benchmarks

MMLU Pro: 52.56
GPQA: 19.24
MUSR: 19.73
BBH: 51.86
IFEval: 38.24 vs 88 (so35)-56.5%
ARC: 68.77 vs 96.7 (so35)-28.9%
HellaSwag: 87.27 vs 95.3 (gpt4)-8.4%
MMLU: 83.79 vs 88.3 (so35)-5.1%
TruthfulQA: 54.74 vs 59 (gpt4)-7.2%
WinoGrande: 84.45 vs 87.5 (gpt4)-3.5%
GSM8K: 85.29 vs 96.4 (so35)-11.5%
MATH Lvl 5: 31.12
LLME Score: 0.32263
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Qwen2 72B (Qwen/Qwen2-72B)
🌟 Advertise your project 🚀

Qwen2 72B Parameters and Internals

Model Type 
text-generation
Additional Notes 
Qwen2 includes a Mixture-of-Experts model and is based on the Transformer architecture with improvements like SwiGLU activation, attention QKV bias, group query attention, and has an adaptive tokenizer for multiple languages and code.
Supported Languages 
en (English)
LLM NameQwen2 72B
Repository 🤗https://huggingface.co/Qwen/Qwen2-72B 
Model Size72b
Required VRAM145.5 GB
Updated2025-09-23
MaintainerQwen
Model Typeqwen2
Model Files  3.8 GB: 1-of-37   4.0 GB: 2-of-37   3.8 GB: 3-of-37   4.0 GB: 4-of-37   4.0 GB: 5-of-37   4.0 GB: 6-of-37   3.8 GB: 7-of-37   4.0 GB: 8-of-37   4.0 GB: 9-of-37   4.0 GB: 10-of-37   3.8 GB: 11-of-37   4.0 GB: 12-of-37   4.0 GB: 13-of-37   4.0 GB: 14-of-37   3.8 GB: 15-of-37   4.0 GB: 16-of-37   4.0 GB: 17-of-37   4.0 GB: 18-of-37   3.8 GB: 19-of-37   4.0 GB: 20-of-37   4.0 GB: 21-of-37   4.0 GB: 22-of-37   3.8 GB: 23-of-37   4.0 GB: 24-of-37   4.0 GB: 25-of-37   4.0 GB: 26-of-37   3.8 GB: 27-of-37   4.0 GB: 28-of-37   4.0 GB: 29-of-37   4.0 GB: 30-of-37   3.8 GB: 31-of-37   4.0 GB: 32-of-37   4.0 GB: 33-of-37   4.0 GB: 34-of-37   3.8 GB: 35-of-37   4.0 GB: 36-of-37   3.5 GB: 37-of-37
Supported Languagesen
Model ArchitectureQwen2ForCausalLM
Licenseother
Context Length131072
Model Max Length131072
Transformers Version4.40.1
Tokenizer ClassQwen2Tokenizer
Padding Token<|endoftext|>
Vocabulary Size152064
Torch Data Typebfloat16
Errorsreplace

Quantized Models of the Qwen2 72B

Model
Likes
Downloads
VRAM
...in 2.9.2 Qwen2 72B 6.0bpw EXL21056 GB
...phin 292 Qwen2 72b EXL2 4 0bpw1438 GB
...n 2.9.2 Qwen72b 8.0bpw H8 EXL22166 GB

Best Alternatives to Qwen2 72B

Best Alternatives
Context / RAM
Downloads
Likes
...R10528DistillQwen 72B Preview3195K / 146 GB156
Kimi Dev 72B128K / 83.4 GB6066367
Qwen2.5 72B128K / 145.5 GB3042478
Ultiima 72B128K / 146.1 GB1303
Homer V1.0 Qwen2.5 72B128K / 146.1 GB656
Ultiima 72B V1.5128K / 146.1 GB240
EVA Qwen2.5 72B V0.2128K / 146 GB18722
AceInstruct 72B128K / 146 GB18417
...n2.5 72B 2x Instruct TIES V1.0128K / 146.1 GB61
Dolphin 2.9.2 Qwen2 72B128K / 146 GB1515162
Note: green Score (e.g. "73.2") means that the model is better than Qwen/Qwen2-72B.

Rank the Qwen2 72B 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? 51545 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