© 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. 8-RES_POST_32K-OAI
    11. 9007
    Prev
    Next
    INDEX
    Explanations

    phrases related to evaluations or categorizations of quality and effectiveness

    oai_token-act-pair · gpt-4o-mini
    Triggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_post_32k-oai
    Configuration
    jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_8.pt
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.8.hook_resid_post
    Hook Layer
    8
    Architecture
    standard
    Context Size
    64
    Dataset
    Skylion007/openwebtext
    Activation Function
    topk
    Embeds
    IFrame
    Link
    Neuron Alignment
    Index
    Value
    % of L₁
    685
    +0.12
    0.6%
    262
    +0.11
    0.5%
    436
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.12
    0.17
    575
    +0.11
    0.14
    724
    +0.09
    0.14
    Negative Logits
    SPONSORED
    -0.61
     parap
    -0.59
    chi
    -0.58
    cientious
    -0.56
    involved
    -0.55
     Flavoring
    -0.55
     Hurricanes
    -0.53
     Canal
    -0.53
     Expedition
    -0.52
     diseng
    -0.51
    POSITIVE LOGITS
     form
    0.93
     shape
    0.79
     apiece
    0.76
    ï¸
    0.72
     Became
    0.70
     avail
    0.70
     forms
    0.70
    format
    0.69
     unden
    0.69
     shapes
    0.69
    Activations Density 3.431%

    No Known Activations