© 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. 31121
    Prev
    Next
    INDEX
    Explanations

    instances of the word "skip" and related terms indicating omission or avoidance

    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
    466
    +0.17
    1.0%
    176
    +0.13
    0.8%
    404
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    466
    +0.17
    0.04
    94
    +0.13
    0.03
    376
    +0.12
    -0.01
    Negative Logits
    iator
    -2.04
    ison
    -1.90
    ity
    -1.62
    ITY
    -1.61
    ised
    -1.55
    eer
    -1.54
    eration
    -1.44
    iance
    -1.43
    illance
    -1.40
     refrig
    -1.39
    POSITIVE LOGITS
    ÅĽÄĩ
    1.98
    blr
    1.70
    ÅĽci
    1.62
     successful
    1.56
    ¹
    1.54
     steps
    1.54
    thood
    1.53
    ĨĴ
    1.52
    ÅĦ
    1.49
    steps
    1.44
    Activations Density 1.061%

    No Known Activations