© 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. 26581
    Prev
    Next
    INDEX
    Explanations

    phrases indicating actions taken without permission or proper protocols

    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
    422
    +0.13
    0.6%
    369
    +0.11
    0.5%
    477
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    234
    +0.13
    0.04
    172
    +0.11
    0.03
    616
    +0.10
    0.04
    Negative Logits
     embodiments
    -0.67
    ibrary
    -0.66
    atures
    -0.64
    outer
    -0.64
    Ranked
    -0.63
    ãĥĺ
    -0.62
    rahim
    -0.61
    alities
    -0.60
    oven
    -0.60
    rawdownloadcloneembedreportprint
    -0.60
    POSITIVE LOGITS
     ever
    0.84
     realizing
    0.84
     warning
    0.83
     hesitation
    0.82
     noticing
    0.82
     provocation
    0.80
     permission
    0.77
     bothering
    0.76
     incident
    0.76
    lication
    0.74
    Activations Density 0.370%

    No Known Activations