What are the hardware requirements for Pythia 31M?

Pythia 31M requires approximately 0.1 GB of VRAM and supports a context window of 2K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Pythia 31M and how large is it?

Pythia 31M is developed by ethzanalytics, a model with 31m parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Pythia 31M perform on standard benchmarks?

Pythia 31M has the following published scores: MMLU 24.27. Compare against reference models on this page or on the LLM Explorer leaderboards.

Pythia 31M by ethzanalytics — VRAM 0.1GB, 2K context

Name: Pythia 31M
Author: ethzanalytics

Pythia 31M is an open-source language model by ethzanalytics. Features: 31m LLM, VRAM: 0.1GB, Context: 2K, License: apache-2.0, LLM Explorer Score: 0.14, Arc: 20, HellaSwag: 26.3, MMLU: 24.3.

Base model:eleutherai/pythia-3... Base model:finetune:eleutherai... Dataset:eleutherai/pile En Endpoints compatible Gpt neox Region:us Safetensors Smol

Model Card on HF 🤗: https://huggingface.co/ethzanalytics/pythia-31m

Pythia 31M Parameters and Internals

Model Type

text-generation

Additional Notes

This is a smaller version of the 'official' checkpoints included in the Pythia study and explicitly saved in fp32 to see safetensors params.

Supported Languages

primary (en)

LLM Name	Pythia 31M
Repository 🤗	https://huggingface.co/ethzanalytics/pythia-31m
Base Model(s)	Pythia 31M EleutherAI/pythia-31m
Model Size	31m
Required VRAM	0.1 GB
Updated	2026-07-27
Maintainer	ethzanalytics
Model Type	gpt_neox
Model Files	0.1 GB
Supported Languages	en
Model Architecture	GPTNeoXForCausalLM
License	apache-2.0
Context Length	2048
Model Max Length	2048
Transformers Version	4.33.1
Tokenizer Class	GPTNeoXTokenizer
Vocabulary Size	50304
Torch Data Type	float32

Rank the Pythia 31M capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

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

Best Alternatives to Pythia 31M

Best Alternatives	Context / RAM	Downloads	Likes
Pythia 31M	2K / 0.1 GB	20717	1
Pythia 31M Chat V1	2K / 0.1 GB	181	8
...implepile Lite 2048 Scratch 2e	2K / 0.1 GB	75	0
...ia 31M Simplewiki Scratch Bf16	2K / 0.1 GB	53	0
... Goodwiki Deduped 2048 Scratch	2K / 0.1 GB	77	1
Pythia 31M Simplewiki 2048	2K / 0.1 GB	51	1
Pythia 31M KI V1 2048 Scratch	2K / 0.1 GB	53	0
...31M SimpleRW Lite 2048 Scratch	2K / 0.1 GB	5	1

Note: green Score (e.g. "73.2") means that the model is better than ethzanalytics/pythia-31m.