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

Pythia 410M Deduped by EleutherAI

 »  All LLMs  »  EleutherAI  »  Pythia 410M Deduped   URL Share it on

  19183   
  Arxiv:2101.00027   Arxiv:2201.07311   Arxiv:2304.01373   Autotrain compatible Dataset:eleutherai/the pile de...   En   Endpoints compatible   Gpt neox   Pythia   Pytorch   Region:us   Safetensors
Model Card on HF 🤗: https://huggingface.co/EleutherAI/pythia-410m-deduped 

Pythia 410M Deduped Benchmarks

ARC: 24.83 vs 96.7 (so35)-74.3%
HellaSwag: 41.29 vs 95.3 (gpt4)-56.7%
MMLU: 25.99 vs 88.3 (so35)-70.6%
TruthfulQA: 40.95 vs 59 (gpt4)-30.6%
WinoGrande: 54.38 vs 87.5 (gpt4)-37.9%
GSM8K: 0.3 vs 96.4 (so35)-99.7%
LLME Score: 0.15656
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Pythia 410M Deduped (EleutherAI/pythia-410m-deduped)
🌟 Advertise your project 🚀

Pythia 410M Deduped Parameters and Internals

Model Type 
Transformer-based Language Model, Causal Language Modeling
Use Cases 
Areas:
Research, Scientific Experiments
Applications:
Interpretability Research
Primary Use Cases:
Analyzing behavior and functionality of large language models
Limitations:
Not suitable for translation or non-English text generation, Not intended for deployment in human-facing interactions
Considerations:
Text generated may be socially unacceptable or undesirable. Users should conduct risk assessments.
Additional Notes 
Model checkpoints are available on Hugging Face hosted as branches for further fine-tuning.
Supported Languages 
languages (.English), proficiency (.High)
Training Details 
Data Sources:
The Pile (globally deduplicated)
Data Volume:
.825 GiB
Model Architecture:
Transformer-based
Safety Evaluation 
Methodologies:
LM Evaluation Harness
Risk Categories:
Misinformation, Bias
Ethical Considerations:
The model is trained on the Pile, which is known to contain profanity and offensive text.
Responsible Ai Considerations 
Fairness:
The Pile contains biases related to gender, religion, and race. Users should conduct their own risk and bias assessments before deployment.
Accountability:
EleutherAI is responsible for the training and release of the model.
Mitigation Strategies:
None provided directly; users are advised to curate model outputs before presentation.
Input Output 
Input Format:
Text input for causal language modeling.
Accepted Modalities:
Text
Output Format:
Text generation as the next token prediction.
Performance Tips:
Fine-tune appropriately; ensure model outputs are curated before use.
Release Notes 
Version:
Current release
Date:
January 2023
Notes:
Renaming of models, retrained with uniform batch sizes and checkpoints.
Version:
Early release
Notes:
Initial release of models with hyperparameter discrepancies.
LLM NamePythia 410M Deduped
Repository 🤗https://huggingface.co/EleutherAI/pythia-410m-deduped 
Model Size410m
Required VRAM0.9 GB
Updated2025-09-23
MaintainerEleutherAI
Model Typegpt_neox
Model Files  0.9 GB   0.9 GB
Supported Languagesen
Model ArchitectureGPTNeoXForCausalLM
Licenseapache-2.0
Context Length2048
Model Max Length2048
Transformers Version4.24.0
Tokenizer ClassGPTNeoXTokenizer
Vocabulary Size50304
Torch Data Typefloat16

Best Alternatives to Pythia 410M Deduped

Best Alternatives
Context / RAM
Downloads
Likes
...thia 410M Cell Type Prediction4K / 0 GB16396
Pythia 410M Sft Full2K / 0.8 GB70
Pythia 410M2K / 0.9 GB3861235
Pythia410m Sft Tldr2K / 1.6 GB11870
Healix 410M2K / 1.6 GB3590
Pythia 410M Deduped SimPOW 02K / 0.8 GB50
Pythia 410M Orpo2K / 1.6 GB50
...7 Kl 01 Steps 12000 Rlhf Model2K / 1.6 GB50
Pythia 410M Ludii Sft2K / 1.6 GB50
... Llm Pythia 410M Pm Gen Ian Nd2K / 1.6 GB50
Note: green Score (e.g. "73.2") means that the model is better than EleutherAI/pythia-410m-deduped.

Rank the Pythia 410M Deduped 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? 51557 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