© 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. 1-RES-SM
    11. 25389
    Prev
    Next
    INDEX
    Explanations

    phrases indicating that something is done accurately

    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/1-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.1.hook_resid_post
    Hook Layer
    1
    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
    376
    +0.13
    0.8%
    148
    +0.11
    0.6%
    377
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    377
    +0.13
    0.02
    78
    +0.11
    0.02
    404
    +0.11
    0.01
    Negative Logits
    ·¸
    -2.12
    ®
    -2.01
    ĥ
    -1.98
    ľ
    -1.91
    Ķ
    -1.89
    Ĭ
    -1.86
    ħ
    -1.82
    icity
    -1.74
    ĨĴ
    -1.74
    Ļ
    -1.73
    POSITIVE LOGITS
     guessed
    1.69
     "%
    1.58
     ellip
    1.57
     enough
    1.56
     @"
    1.42
    eu
    1.40
     equip
    1.38
     "$(
    1.37
     speaking
    1.36
     None
    1.35
    Activations Density 0.035%

    No Known Activations