© 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. 2-RES-SM
    11. 26291
    Prev
    Next
    INDEX
    Explanations

    references to specific software or programming environments

    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/2-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.2.hook_resid_post
    Hook Layer
    2
    Architecture
    standard
    Context Size
    128
    Dataset
    EleutherAI/the_pile_deduplicated
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Neuron Alignment
    Index
    Value
    % of L₁
    368
    +0.16
    0.9%
    195
    +0.13
    0.7%
    395
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    58
    +0.16
    0.02
    252
    +0.13
    0.02
    368
    +0.11
    0.01
    Negative Logits
     tight
    -1.45
    iliated
    -1.42
     anything
    -1.38
    ay
    -1.37
     cron
    -1.36
    ming
    -1.34
    med
    -1.34
    ayed
    -1.33
    withstanding
    -1.32
    ned
    -1.31
    POSITIVE LOGITS
    ®
    2.73
    ĥ
    2.62
    Ĩ
    2.39
    ĩ
    2.30
    ¢
    2.27
    ī
    2.22
    º
    2.19
    ¹
    2.19
    ¿
    2.18
    ¡
    2.16
    Activations Density 0.120%

    No Known Activations