© 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. 30640
    Prev
    Next
    INDEX
    Explanations

    closing braces and code structure indicators

    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
    415
    +0.12
    0.7%
    9
    +0.12
    0.7%
    360
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    360
    +0.12
    0.06
    46
    +0.12
    0.05
    415
    +0.11
    0.05
    Negative Logits
    )<
    -1.51
     punishment
    -1.45
    odor
    -1.41
    )].
    -1.40
    free
    -1.40
    amines
    -1.39
    )](
    -1.36
     deserves
    -1.31
    sten
    -1.28
    blank
    -1.28
    POSITIVE LOGITS
     Then
    1.63
     finally
    1.56
     then
    1.52
    «
    1.49
     else
    1.46
    ¦
    1.45
    ĻĤ
    1.40
    ģ
    1.38
     maj
    1.34
    ĭ
    1.33
    Activations Density 0.216%

    No Known Activations