© 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. 2-RES-SM
    11. 31979
    Prev
    Next
    INDEX
    Explanations

    phrases indicating actions or expectations directed towards others

    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/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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    136
    +0.16
    0.9%
    69
    +0.16
    0.9%
    142
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    69
    +0.16
    0.13
    505
    +0.16
    0.10
    406
    +0.13
    0.09
    Negative Logits
    er
    -1.79
     glimps
    -1.59
    ered
    -1.59
    eria
    -1.50
     hairs
    -1.48
    arf
    -1.45
    hest
    -1.43
     goodbye
    -1.40
    ty
    -1.39
    cap
    -1.38
    POSITIVE LOGITS
    asts
    2.15
     pursue
    1.56
    ductive
    1.54
     respective
    1.54
     include
    1.48
    erce
    1.45
    rapeut
    1.43
     occupy
    1.42
    endered
    1.41
    missions
    1.39
    Activations Density 0.728%

    No Known Activations