© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 9246
    Prev
    Next
    INDEX
    Explanations

    punctuation, specifically commas

    oai_token-act-pair · gpt-4o-miniTriggered 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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    21
    +0.16
    1.0%
    78
    +0.15
    0.9%
    203
    +0.11
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    391
    +0.16
    0.02
    191
    +0.15
    0.02
    114
    +0.11
    0.01
    Negative Logits
    abine
    -1.54
     around
    -1.52
    ogens
    -1.51
    izing
    -1.49
     goodbye
    -1.47
    rier
    -1.45
    pad
    -1.43
    imals
    -1.41
    helial
    -1.40
    izers
    -1.40
    POSITIVE LOGITS
     Redistributions
    1.81
     Version
    1.73
    agreement
    1.55
    INCLUDING
    1.55
    fileID
    1.41
    LICENSE
    1.41
    об
    1.38
    distributed
    1.33
     MERCHANTABILITY
    1.31
    Ñģли
    1.30
    Activations Density 0.027%

    No Known Activations