© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW 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. 26485
    Prev
    Next
    INDEX
    Explanations

    punctuations and pauses in dialogue

    oai_token-act-pair · gpt-4o-mini
    Triggered 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
    Neuron Alignment
    Index
    Value
    % of L₁
    314
    +0.23
    1.0%
    373
    +0.11
    0.5%
    160
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    720
    +0.23
    0.05
    267
    +0.11
    0.04
    100
    +0.10
    0.03
    Negative Logits
     deserts
    -0.74
     distribut
    -0.73
     interchange
    -0.71
     sanctioned
    -0.70
     fielded
    -0.70
     merit
    -0.69
     parap
    -0.69
     materials
    -0.68
     fict
    -0.68
     directional
    -0.68
    POSITIVE LOGITS
    why
    1.43
    they
    1.42
    there
    1.41
    then
    1.38
    you
    1.37
    but
    1.36
    maybe
    1.35
    what
    1.35
    this
    1.32
    yet
    1.32
    Activations Density 0.101%

    No Known Activations