© 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. 8990
    Prev
    Next
    INDEX
    Explanations

    phrases that structure or attribute responsibility to different parties or entities

    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
    625
    +0.10
    0.5%
    166
    +0.09
    0.4%
    80
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    703
    +0.10
    0.07
    393
    +0.09
    0.10
    484
    +0.09
    0.09
    Negative Logits
    agues
    -0.70
    Offline
    -0.64
     Offline
    -0.60
     Bought
    -0.59
    ðŁĺ
    -0.58
     sts
    -0.56
     cheapest
    -0.56
     err
    -0.56
     Runs
    -0.54
    anka
    -0.54
    POSITIVE LOGITS
     consequ
    1.04
    particularly
    0.98
     ensuing
    0.94
     particularly
    0.93
     resultant
    0.91
     subsequent
    0.88
     its
    0.86
    especially
    0.81
     implications
    0.80
     especially
    0.80
    Activations Density 0.884%

    No Known Activations