© 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. 8976
    Prev
    Next
    INDEX
    Explanations

    numerical references or citations within the text

    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
    640
    +0.17
    0.7%
    605
    +0.12
    0.5%
    497
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    6
    +0.17
    0.10
    640
    +0.12
    0.08
    679
    +0.10
    0.07
    Negative Logits
     therefore
    -0.72
     Takeru
    -0.71
     nonex
    -0.68
     burner
    -0.66
     UD
    -0.64
     conceivable
    -0.63
     antit
    -0.63
     imaginable
    -0.62
     Tanaka
    -0.61
     inward
    -0.60
    POSITIVE LOGITS
     Alternatively
    1.10
    Alternatively
    0.91
    <|endoftext|>
    0.91
     Also
    0.89
    Or
    0.84
     Includes
    0.80
     Or
    0.79
    FAQ
    0.77
    VID
    0.77
    pdf
    0.77
    Activations Density 0.637%

    No Known Activations