© 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. 8-RES_POST_32K-OAI
    11. 8678
    Prev
    Next
    INDEX
    Explanations

    references to Saudi Arabia and its cities or associated figures

    oai_token-act-pair · gpt-4o-miniTriggered 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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    311
    +0.12
    0.5%
    143
    +0.10
    0.4%
    288
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.12
    0.02
    57
    +0.10
    0.02
    146
    +0.09
    0.02
    Negative Logits
     Puzz
    -0.77
    acht
    -0.74
    schild
    -0.72
    lus
    -0.68
    ombie
    -0.66
    inez
    -0.64
     Surviv
    -0.63
    WER
    -0.62
    Ñı
    -0.62
     Niet
    -0.61
    POSITIVE LOGITS
     Arabia
    1.84
     Arabian
    1.57
     Aram
    1.06
    ishi
    0.80
    imaru
    0.78
     Abdullah
    0.73
     Tayyip
    0.73
    Ļ½
    0.71
     Gazette
    0.71
     Abdul
    0.70
    Activations Density 0.069%

    No Known Activations