© 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. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    5. GPT2-Small
    7. Residual Post - 32k
    9. 10-RES_POST_32K-OAI
    11. 30804
    Prev
    Next
    INDEX
    Explanations

    instances of ellipses or interruptions in narrative flow

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 10-res_post_32k-oai
    Configuration
    jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_10.pt
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.10.hook_resid_post
    Hook Layer
    10
    Architecture
    standard
    Context Size
    64
    Dataset
    Skylion007/openwebtext
    Activation Function
    topk
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    65
    +0.10
    0.5%
    607
    +0.09
    0.4%
    562
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    406
    +0.10
    0.05
    675
    +0.09
    0.04
    457
    +0.09
    0.04
    Negative Logits
    Therefore
    -0.77
     and
    -0.65
    atche
    -0.65
     hence
    -0.65
     Chero
    -0.64
    icas
    -0.64
     Indra
    -0.64
     whereas
    -0.64
    SourceFile
    -0.63
     Drawn
    -0.63
    POSITIVE LOGITS
     uh
    1.19
     um
    1.08
    uh
    0.89
    well
    0.86
     err
    0.83
     er
    0.79
     unconventional
    0.77
     gasp
    0.75
    wait
    0.74
    different
    0.73
    Activations Density 0.620%

    No Known Activations