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

Pythia 12B by EleutherAI

 »  All LLMs  »  EleutherAI  »  Pythia 12B   URL Share it on

  20051   

Pythia 12B is an open-source language model by EleutherAI. Features: 12b LLM, VRAM: 23.8GB, Context: 2K, License: apache-2.0, LLM Explorer Score: 0.16, Arc: 39.6, HellaSwag: 68.8, MMLU: 26.8, GSM8K: 1.7.

  Arxiv:2101.00027   Arxiv:2201.07311   Arxiv:2304.01373   Dataset:eleutherai/pile   Deploy:azure   En   Endpoints compatible   Gpt neox   Pythia   Pytorch   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/EleutherAI/pythia-12b 

Pythia 12B Benchmarks

ARC: 39.59 vs 96.7 (so35)-59.1%
HellaSwag: 68.82 vs 95.3 (gpt4)-27.8%
MMLU: 26.76 vs 88.3 (so35)-69.7%
GSM8K: 1.74 vs 96.4 (so35)-98.2%
LLME Score: 0.15931
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Pythia 12B (EleutherAI/pythia-12b)
🌟 Advertise your project 🚀

Pythia 12B Parameters and Internals

Model Type 
Transformer-based Language Model
Use Cases 
Areas:
Research
Applications:
Scientific experiments
Primary Use Cases:
Interpretability Research
Limitations:
Not intended for deployment, English-language only, Not suitable for translation or generating text in other languages
Considerations:
Potential for generating harmful or offensive text
Additional Notes 
Pythia models are intended to facilitate interpretability research. They are not suitable as standalone products for human interactions.
Supported Languages 
English (fluent)
Training Details 
Data Sources:
EleutherAI/pile
Data Volume:
299,892,736,000 tokens
Methodology:
Training for 143000 steps, batch size 2M tokens
Model Architecture:
Transformer-based
Input Output 
Input Format:
String of text
Accepted Modalities:
Text
Output Format:
String of text
Release Notes 
Version:
Current release
Date:
January 2023
Notes:
Renamed Pythia models and retrained addressing hyperparameter discrepancies. Uniform batch size for all models.
LLM NamePythia 12B
Repository 🤗https://huggingface.co/EleutherAI/pythia-12b 
Model Size12b
Required VRAM23.8 GB
Updated2026-05-12
MaintainerEleutherAI
Model Typegpt_neox
Model Files  9.8 GB: 1-of-3   9.9 GB: 2-of-3   4.1 GB: 3-of-3   9.8 GB: 1-of-3   9.9 GB: 2-of-3   4.1 GB: 3-of-3
Supported Languagesen
Model ArchitectureGPTNeoXForCausalLM
Licenseapache-2.0
Context Length2048
Model Max Length2048
Transformers Version4.24.0
Tokenizer ClassGPTNeoXTokenizer
Vocabulary Size50688
Torch Data Typefloat16

Best Alternatives to Pythia 12B

Best Alternatives
Context / RAM
Downloads
Likes
Dolly V2 12B2K / 23.8 GB32831955
...sst Sft 4 Pythia 12B Epoch 3.52K / 23.8 GB2058368
Oasst Rl 1 Pythia 12B2K / 23.8 GB94
Oasst Sft 1 Pythia 12B2K / 23.8 GB1231278
H2ogpt Gm Oasst1 En 1024 12B2K / 23.8 GB18835
Pythia 12B Deduped2K / 23.8 GB338252
Pythia 12B Sft V8 7K Steps2K / 23.8 GB110622
...ythia 12B Sft V8 Rlhf 2K Steps2K / 23.8 GB10010
H2ogpt Oasst1 512 12B2K / 23.9 GB112929
Pythia 12B Sft V8.2.5K Steps2K / 23.8 GB10280
Note: green Score (e.g. "73.2") means that the model is better than EleutherAI/pythia-12b.

Rank the Pythia 12B 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? 53999 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