par EleutherAI
Open source · 133k downloads · 1 likes
Pythia-14M est un petit modèle de langage basé sur une architecture Transformer, développé par EleutherAI pour la recherche en intelligence artificielle. Conçu comme un outil d'étude, il permet d'analyser le comportement, les mécanismes et les limites des grands modèles de langage dans un cadre contrôlé. Bien que capable de générer du texte en anglais, il n'est pas optimisé pour des applications pratiques comme les chatbots ou la rédaction professionnelle, car il n'a pas subi de fine-tuning spécifique. Son principal atout réside dans sa transparence : il offre 154 points de contrôle (checkpoints) accessibles, permettant d'étudier son évolution au fil de l'entraînement. Cependant, ses réponses peuvent être biaisées, inexactes ou socialement inappropriées, en raison des données d'entraînement utilisées, et il ne doit pas être utilisé pour des interactions critiques ou sans supervision humaine.
#release-discussion. For general correspondence: [email protected]Note: Prior to Feb 27th, 2026, a model was hosted at this URL that was trained on the deduplicated Pile. That model is now at EleutherAI/pythia-14m-deduped and this model was correctly trained on the standard Pile. We apologize for any confusion this has caused.
Note: Pythia-14M was trained after the original Pythia suite at the request of interpretability researchers who wanted a smaller model with the same training setup. It uses the same tokenizer, hyperparameter conventions, and checkpoint schedule as the rest of the Pythia suite. A deduplicated variant is available at
EleutherAI/pythia-14m-deduped.
Model Configuration:
| Pythia model | Non-Embedding Params | Layers | Model Dim | Heads | Batch Size | Learning Rate | Equivalent Models |
|---|---|---|---|---|---|---|---|
| 14M | 1,189,888 | 6 | 128 | 4 | 2M | 1.0 x 10-3 | — |
| 31M | 4,739,072 | 6 | 256 | 8 | 2M | 1.0 x 10-3 | — |
| 70M | 18,915,328 | 6 | 512 | 8 | 2M | 1.0 x 10-3 | — |
| 160M | 85,056,000 | 12 | 768 | 12 | 2M | 6.0 x 10-4 | GPT-Neo 125M, OPT-125M |
| 410M | 302,311,424 | 24 | 1024 | 16 | 2M | 3.0 x 10-4 | OPT-350M |
| 1.0B | 805,736,448 | 16 | 2048 | 8 | 2M | 3.0 x 10-4 | — |
| 1.4B | 1,208,602,624 | 24 | 2048 | 16 | 2M | 2.0 x 10-4 | GPT-Neo 1.3B, OPT-1.3B |
| 2.8B | 2,517,652,480 | 32 | 2560 | 32 | 2M | 1.6 x 10-4 | GPT-Neo 2.7B, OPT-2.7B |
| 6.9B | 6,444,163,072 | 32 | 4096 | 32 | 2M | 1.2 x 10-4 | OPT-6.7B |
| 12B | 11,327,027,200 | 36 | 5120 | 40 | 2M | 1.2 x 10-4 | — |
The primary intended use of Pythia is research on the behavior, functionality, and limitations of large language models. This suite is intended to provide a controlled setting for performing scientific experiments. We provide 154 checkpoints per model:
step0step{1,2,4...512}step1000 to step143000These checkpoints are hosted on Hugging Face as branches. Note that branch 143000 corresponds exactly to the model checkpoint on the main branch of each model.
You may further fine-tune and adapt Pythia-14M for deployment, as long as your use is in accordance with the Apache 2.0 license. Pythia models work with the Hugging Face Transformers Library. If you use pre-trained Pythia-14M as a basis for your fine-tuned model, please conduct your own risk and bias assessment.
The Pythia Suite is not intended for deployment. It is not a product and cannot be used for human-facing interactions. For example, the model may generate harmful or offensive text. Please evaluate the risks associated with your particular use case.
Pythia models are English-language only and are not suitable for translation or generating text in other languages.
Pythia-14M has not been fine-tuned for downstream contexts in which language models are commonly deployed, such as writing genre prose or commercial chatbots. This means Pythia-14M will not respond to a given prompt the way a product like ChatGPT does. ChatGPT was fine-tuned using methods such as Reinforcement Learning from Human Feedback (RLHF) to better "follow" human instructions.
The core functionality of a large language model is to take a string of text and predict the next token. The token used by the model need not produce the most "accurate" text. Never rely on Pythia-14M to produce factually accurate output.
This model was trained on the Pile, a dataset known to contain profanity and texts that are lewd or otherwise offensive. See Section 6 of the Pile paper for a discussion of documented biases with regards to gender, religion, and race. Pythia-14M may produce socially unacceptable or undesirable text, even if the prompt itself does not include anything explicitly offensive.
If you plan on using text generated through, for example, the Hosted Inference API, we recommend having a human curate the outputs of this language model before presenting it to other people. Please inform your audience that the text was generated by Pythia-14M.
Pythia models can be loaded and used via the following code, demonstrated here for the third pythia-14m checkpoint:
from transformers import GPTNeoXForCausalLM, AutoTokenizer
model = GPTNeoXForCausalLM.from_pretrained(
"EleutherAI/pythia-14m",
revision="step3000",
cache_dir="./pythia-14m/step3000",
)
tokenizer = AutoTokenizer.from_pretrained(
"EleutherAI/pythia-14m",
revision="step3000",
cache_dir="./pythia-14m/step3000",
)
inputs = tokenizer("Hello, I am", return_tensors="pt")
tokens = model.generate(**inputs)
tokenizer.decode(tokens[0])
Revision/branch step143000 corresponds exactly to the model checkpoint on the main branch of each model.
For more information on how to use all Pythia models, see documentation on GitHub.
The Pile is a 825GiB general-purpose dataset in English. It was created by EleutherAI specifically for training large language models. It contains texts from 22 diverse sources, roughly broken down into five categories:
See the Pile paper for a breakdown of all data sources, methodology, and ethical implications. Consult the datasheet for more detailed documentation about the Pile and its component datasets.
The Pile can be downloaded from the official website or from a community mirror.
Pythia-14M was trained on the Pile.
All models were trained on the exact same data, in the exact same order. Each model saw 299,892,736,000 tokens during training, and 143 checkpoints for each model are saved every 2,097,152,000 tokens, spaced evenly throughout training, from step1000 to step143000 (which is the same as main). Early checkpoints are also provided: step0 and step{1,2,4...512}.
All Pythia models trained for 143,000 steps at a batch size of 2M (2,097,152 tokens).
See GitHub for more details on training procedure, including how to reproduce it.
Pythia uses the same tokenizer as GPT-NeoX-20B.
All 16 Pythia models were evaluated using the LM Evaluation Harness. You can access the results by model and step at results/json/* in the GitHub repository.
This section compares differences between previously released Pythia v0 and the current models. See Appendix B of the Pythia paper for further discussion of these changes and the motivation behind them. We found that retraining Pythia had no impact on benchmark performance.
Pythia models were renamed in January 2023. It is possible that the old naming convention still persists in some documentation by accident. The current naming convention (70M, 160M, etc.) is based on total parameter count.
| Current Pythia suffix | Old suffix | Total params | Non-embedding params |
|---|---|---|---|
| 14M | — | 14,067,712 | 1,189,888 |
| 31M | — | 30,494,720 | 4,739,072 |
| 70M | 19M | 70,426,624 | 18,915,328 |
| 160M | 125M | 162,322,944 | 85,056,000 |
| 410M | 350M | 405,334,016 | 302,311,424 |
| 1B | 800M | 1,011,781,632 | 805,736,448 |
| 1.4B | 1.3B | 1,414,647,808 | 1,208,602,624 |
| 2.8B | 2.7B | 2,775,208,960 | 2,517,652,480 |
| 6.9B | 6.7B | 6,857,302,016 | 6,444,163,072 |
| 12B | 13B | 11,846,072,320 | 11,327,027,200 |
@article{biderman2023pythia,
title={Pythia: A Suite for Analyzing Large Language Models Across Training and Scaling},
author={Biderman, Stella and Schoelkopf, Hailey and Anthony, Quentin Gregory and Bradley, Herbie and O'Brien, Kyle and Hallahan, Eric and Khan, Mohammad Aflah and Purohit, Shivanshu and Prashanth, USVSN Sai and Raff, Edward and others},
journal={International Conference on Machine Learning},
year={2023}
}