© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 4-RES-SM
    11. 8704
    Prev
    Next
    INDEX
    Explanations

    references to data models and their validation processes

    oai_token-act-pair · gpt-4o-mini
    Triggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    ctigges/pythia-70m-deduped__res-sm_processed/4-res-sm
    Prompts (Dashboard)
    32,768 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.4.hook_resid_post
    Hook Layer
    4
    Architecture
    standard
    Context Size
    128
    Dataset
    EleutherAI/the_pile_deduplicated
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Neuron Alignment
    Index
    Value
    % of L₁
    307
    +0.15
    0.9%
    156
    +0.13
    0.7%
    17
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    17
    +0.15
    0.03
    255
    +0.13
    -0.01
    370
    +0.12
    0.06
    Negative Logits
    published
    -1.58
    sworth
    -1.53
     canals
    -1.52
    ashes
    -1.47
    fulness
    -1.43
     coronavirus
    -1.40
    day
    -1.40
    polarized
    -1.38
    idays
    -1.36
     tomatoes
    -1.35
    POSITIVE LOGITS
    Attribute
    1.91
    Framework
    1.89
    Serializer
    1.80
    ERTY
    1.76
    ei
    1.73
    EventListener
    1.73
    Metadata
    1.69
    Resolver
    1.69
    yset
    1.65
    Instance
    1.65
    Activations Density 0.404%

    No Known Activations