© 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 EvalsExportsinterp-engineNEWAPI 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. 27377
    Prev
    Next
    INDEX
    Explanations

    references to controversial or provocative subjects, particularly involving personal experiences or societal commentary

    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
    212
    +0.12
    0.6%
    53
    +0.11
    0.5%
    39
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    351
    +0.12
    0.18
    762
    +0.11
    0.21
    241
    +0.10
    0.15
    Negative Logits
     undertook
    -0.79
     undertaken
    -0.78
     util
    -0.74
    isSpecialOrderable
    -0.69
     locality
    -0.68
     desired
    -0.68
     envis
    -0.68
     complementary
    -0.67
     elaborated
    -0.67
     recomm
    -0.67
    POSITIVE LOGITS
     lol
    1.36
     haha
    1.27
     anymore
    1.27
     crap
    1.25
     shit
    1.22
     LOL
    1.19
     dude
    1.18
     someday
    1.15
     bullshit
    1.15
    ?!
    1.13
    Activations Density 3.657%

    No Known Activations