© 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. 8-RES_POST_32K-OAI
    11. 8927
    Prev
    Next
    INDEX
    Explanations

    numerical values or references to statistics

    oai_token-act-pair · gpt-4o-mini
    Triggered 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
    Neuron Alignment
    Index
    Value
    % of L₁
    679
    +0.10
    0.5%
    547
    +0.10
    0.5%
    221
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    718
    +0.10
    0.22
    195
    +0.10
    0.04
    540
    +0.10
    0.21
    Negative Logits
    enery
    -0.73
     Tycoon
    -0.71
     occupancy
    -0.70
     Mankind
    -0.68
     renovations
    -0.68
    iage
    -0.65
     Canaveral
    -0.64
    Virgin
    -0.63
    yrim
    -0.63
    thood
    -0.63
    POSITIVE LOGITS
     rhetorical
    1.00
     disingen
    0.98
     rhet
    0.97
     methodological
    0.97
     empir
    0.89
     miscon
    0.88
     cynicism
    0.87
     Chomsky
    0.86
     analytic
    0.85
     clar
    0.85
    Activations Density 9.920%

    No Known Activations