© 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. 32181
    Prev
    Next
    INDEX
    Explanations

    references to legal codes and sections

    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
    314
    +0.13
    0.6%
    749
    +0.11
    0.5%
    96
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    314
    +0.13
    -0.00
    466
    +0.11
    0.03
    103
    +0.10
    0.02
    Negative Logits
     Reborn
    -0.67
    ipeg
    -0.67
    heastern
    -0.66
    neys
    -0.65
     dexter
    -0.63
     radiant
    -0.63
    etsk
    -0.62
     congratulations
    -0.62
    ariat
    -0.62
     sooner
    -0.61
    POSITIVE LOGITS
    552
    0.89
     702
    0.88
    1983
    0.85
     1915
    0.82
     504
    0.82
     1070
    0.82
    Title
    0.81
    Act
    0.81
     802
    0.80
    441
    0.80
    Activations Density 0.190%

    No Known Activations