© 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. 8635
    Prev
    Next
    INDEX
    Explanations

    references to academic publications and citations

    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
    156
    +0.17
    1.0%
    466
    +0.14
    0.8%
    168
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    168
    +0.17
    0.07
    203
    +0.14
    0.04
    266
    +0.12
    0.05
    Negative Logits
    ĻĤ
    -3.96
    ¢
    -3.23
    ª
    -3.19
    ¿½
    -3.16
    ±
    -3.14
    ©
    -3.11
    »
    -3.10
    ´
    -3.10
    Ń
    -3.06
    Ĺ
    -3.06
    POSITIVE LOGITS
     Inc
    1.98
     Department
    1.96
     University
    1.93
     Bureau
    1.85
     Dept
    1.85
     Ltd
    1.79
     któ
    1.78
     Faculty
    1.66
     courtesy
    1.66
    which
    1.66
    Activations Density 0.432%

    No Known Activations