Skip to content

@EleutherAI EleutherAI

Change the repository type filter

All

    Repositories list

    153 repositories

    • lm-evaluation-harness

      Public
      A framework for few-shot evaluation of language models.
      transformerlanguage-modelevaluation-framework
      Python
      •
      MIT License
      •1.9k•7.1k•322•93•Updated Dec 4, 2024Dec 4, 2024

    • ngrams-across-time

      Public
      Jupyter Notebook
      •
      MIT License
      •0•4•0•0•Updated Dec 4, 2024Dec 4, 2024

    • transformer-reasoning

      Public
      Experiments in transformer knowledge and reasoning
      Jupyter Notebook
      •
      MIT License
      •12•3•0•0•Updated Dec 4, 2024Dec 4, 2024

    • sae

      Public
      Sparse autoencoders
      Python
      •
      MIT License
      •50•362•5•2•Updated Dec 4, 2024Dec 4, 2024

    • gpt-neox

      Public
      An implementation of model parallel autoregressive transformers on GPUs, based on the Megatron and DeepSpeed libraries
      transformerslanguage-modelgpt-3deepspeed-library
      Python
      •
      Apache License 2.0
      •1k•7k•59•24•Updated Dec 3, 2024Dec 3, 2024

    • aria-utils

      Public
      MIDI tokenizers and pre-processing utils.
      Python
      •
      Apache License 2.0
      •1•0•0•0•Updated Dec 3, 2024Dec 3, 2024

    • elk

      Public
      Keeping language models honest by directly eliciting knowledge encoded in their activations.
      Python
      •
      MIT License
      •33•187•15•10•Updated Dec 2, 2024Dec 2, 2024

    • aria-amt

      Public
      Efficient and robust implementation of seq-to-seq automatic piano transcription.
      Python
      •
      Apache License 2.0
      •8•32•0•0•Updated Dec 2, 2024Dec 2, 2024

    • basin-volume

      Public
      Precisely estimating the volume of basins in neural net parameter space corresponding to interpretable behaviors
      Jupyter Notebook
      •
      Apache License 2.0
      •0•0•0•0•Updated Nov 30, 2024Nov 30, 2024

    • mdl

      Public
      Minimum Description Length probing for neural network representations
      Python
      •
      MIT License
      •2•16•0•2•Updated Nov 23, 2024Nov 23, 2024

    • polyapprox

      Public
      Closed-form polynomial approximations to neural networks
      Python
      •
      MIT License
      •0•2•0•0•Updated Nov 22, 2024Nov 22, 2024

    • website

      Public
      New website for EleutherAI based on Hugo static site generator
      HTML
      •6•4•0•2•Updated Nov 21, 2024Nov 21, 2024

    • nanoGPT-mup

      Public
      The simplest, fastest repository for training/finetuning medium-sized GPTs.
      Python
      •
      MIT License
      •6k•85•0•0•Updated Nov 19, 2024Nov 19, 2024

    • sae-auto-interp

      Public
      Jupyter Notebook
      •
      Apache License 2.0
      •12•118•6•1•Updated Nov 19, 2024Nov 19, 2024

    • semantic-memorization

      Public
      Jupyter Notebook
      •5•46•1•5•Updated Nov 17, 2024Nov 17, 2024

    • concept-erasure

      Public
      Erasing concepts from neural representations with provable guarantees
      Python
      •
      MIT License
      •15•213•2•2•Updated Nov 13, 2024Nov 13, 2024

    • monkfish

      Public
      Python
      •
      MIT License
      •1•4•0•0•Updated Nov 1, 2024Nov 1, 2024

    • pythia

      Public
      The hub for EleutherAI's work on interpretability and learning dynamics
      Jupyter Notebook
      •
      Apache License 2.0
      •173•2.3k•25•3•Updated Nov 1, 2024Nov 1, 2024

    • features-across-time

      Public
      Understanding how features learned by neural networks evolve throughout training
      Python
      •
      MIT License
      •1•31•0•0•Updated Oct 24, 2024Oct 24, 2024

    • scalable-elicitation

      Public
      The code used in "Balancing Label Quantity and Quality for Scalable Elicitation"
      Jupyter Notebook
      •
      MIT License
      •3•3•0•0•Updated Oct 22, 2024Oct 22, 2024

    • aria

      Public
      Python
      •
      Apache License 2.0
      •11•42•0•0•Updated Oct 9, 2024Oct 9, 2024

    • tokengrams

      Public
      Efficiently computing & storing token n-grams from large corpora
      Rust
      •
      MIT License
      •3•16•0•0•Updated Oct 6, 2024Oct 6, 2024

    • optax-galore

      Public
      Adds GaLore style projection wrappers to optax optimizers
      Python
      •
      MIT License
      •0•4•0•0•Updated Oct 3, 2024Oct 3, 2024

    • equinox-llama

      Public
      Equinox implementation of llama3 and llama3.1
      Python
      •
      MIT License
      •0•7•1•0•Updated Oct 3, 2024Oct 3, 2024

    • cupbearer

      Public
      A library for mechanistic anomaly detection
      Python
      •
      MIT License
      •10•6•0•0•Updated Oct 3, 2024Oct 3, 2024

    • cookbook

      Public
      Deep learning for dummies. All the practical details and useful utilities that go into working with real models.
      Python
      •
      Apache License 2.0
      •38•729•8•0•Updated Sep 24, 2024Sep 24, 2024

    • w2s

      Public
      Python
      •
      MIT License
      •3•18•2•0•Updated Sep 24, 2024Sep 24, 2024

    • ccs

      Public
      Python
      •
      MIT License
      •6•5•3•3•Updated Sep 24, 2024Sep 24, 2024

    • auto-circuit

      Public
      A library for efficient patching and automatic circuit discovery.
      Python
      •13•1•0•0•Updated Sep 16, 2024Sep 16, 2024

    • steering-llama3

      Public
      Python
      •1•18•0•1•Updated Aug 2, 2024Aug 2, 2024