© 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. 31664
    Prev
    Next
    INDEX
    Explanations

    numeric data and references to studies or publications

    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
    6
    +0.14
    0.7%
    430
    +0.14
    0.7%
    314
    +0.14
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    160
    +0.14
    0.04
    107
    +0.14
    0.04
    6
    +0.14
    0.04
    Negative Logits
     confir
    -0.94
     incent
    -0.92
     destro
    -0.91
     distingu
    -0.90
     eleph
    -0.89
     conclud
    -0.89
     princ
    -0.88
     answ
    -0.85
     occas
    -0.80
     recip
    -0.79
    POSITIVE LOGITS
    Hong
    1.23
    Xi
    1.21
    PRES
    1.20
    SAN
    1.20
    Naz
    1.18
    Yan
    1.15
    Na
    1.13
    Yo
    1.11
    FACE
    1.11
    Wan
    1.10
    Activations Density 0.112%

    No Known Activations