© 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 EvalsExportsAPI 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. 10646
    Prev
    Next
    INDEX
    Explanations

    mention of the name "Will" in various contexts

    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
    156
    +0.16
    0.9%
    410
    +0.15
    0.8%
    369
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    410
    +0.16
    0.04
    202
    +0.15
    0.03
    415
    +0.13
    0.02
    Negative Logits
    unnumbered
    -1.96
    »¿
    -1.89
    į
    -1.81
    matically
    -1.60
    ĥ
    -1.55
    ĩ
    -1.54
    Ļª
    -1.52
    ĥ½
    -1.51
    ypse
    -1.50
    anced
    -1.50
    POSITIVE LOGITS
    ful
    2.34
    ough
    2.04
    cox
    1.80
    fulness
    1.76
    fully
    1.72
    iger
    1.60
    zier
    1.49
    shire
    1.48
     petition
    1.47
    bell
    1.43
    Activations Density 0.225%

    No Known Activations