LLM EXPLORER 59,420 MODELS INDEXED

Pythia 410M by EleutherAI

By EleutherAI · 261700 downloads

Pythia 410M is an open-source language model by EleutherAI. Features: 410m LLM, VRAM: 0.9GB, Context: 2K, License: apache-2.0, LLM Explorer Score: 0.19, Arc: 26.2, HellaSwag: 40.9, MMLU: 27.3, GSM8K: 0.7.

  Arxiv:2101.00027   Arxiv:2201.07311   Arxiv:2304.01373   Dataset:eleutherai/pile   Deploy:azure   En   Endpoints compatible   Gpt neox   Pythia   Pytorch   Region:us   Safetensors
Model Card on HF πŸ€—: https://huggingface.co/EleutherAI/pythia-410m 

Pythia 410M Parameters and Internals

Model Type 
Transformer-based Language Model
Use Cases 
Areas:
Research
Primary Use Cases:
Behavior and functionality research of large language models
Limitations:
Not suitable for human-facing deployment, translation or generating text in other languages
Considerations:
Conduct risk and bias assessments when using in downstream applications.
Additional Notes 
Pythia-410M is not tuned for downstream applications like commercial chatbots.
Supported Languages 
en (Primary language - English)
Training Details 
Data Sources:
The Pile
Data Volume:
299,892,736,000 tokens
Methodology:
Trained with uniform batch size of 2M tokens. Used Flash Attention. Learning rate schedule decayed to a minimum of 0.1Γ— maximum LR.
Training Time:
143000 steps at a batch size of 2M
Model Architecture:
Transformer-based
Responsible Ai Considerations 
Fairness:
Biases regarding gender, religion, and race documented in Section 6 of the Pile paper.
Transparency:
Model outputs should not be relied upon for factual accuracy.
Accountability:
Users responsible for evaluating and informing audiences about generated outputs.
Mitigation Strategies:
Implement risk and bias assessments when using in downstream applications.
Input Output 
Input Format:
Text string
Accepted Modalities:
text
Output Format:
Text string
Performance Tips:
Always evaluate the outputs for factual accuracy and potential biases.
Release Notes 
Date:
January 2023
Notes:
Pythia models were renamed and parameter counts adjusted for clarity.
Version:
Pythia v0
Notes:
Early version with hyperparameter discrepancies.
LLM NamePythia 410M
Repository πŸ€—https://huggingface.co/EleutherAI/pythia-410m 
Model Size410m
Required VRAM0.9 GB
Updated2026-08-05
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

Best Alternatives
Context / RAM
Downloads
Likes
...thia 410M Cell Type Prediction4K / 0 GB36610
Pythia 410M Sft Full2K / 0.8 GB41
Pythia 410M Deduped2K / 0.9 GB6259722
Pythia410m Sft Tldr2K / 1.6 GB71590
Healix 410M2K / 1.6 GB880
Pythia 410M Deduped SimPOW 02K / 0.8 GB50
Pythia 410M Orpo2K / 1.6 GB50
...7 Kl 01 Steps 12000 Rlhf Model2K / 1.6 GB80
Pythia 410M Ludii Sft2K / 1.6 GB70
... Llm Pythia 410M Pm Gen Ian Nd2K / 1.6 GB50