© 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. 8948
    Prev
    Next
    INDEX
    Explanations

    instances of business and company designations, particularly "LLC" and "Inc."

    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
    430
    +0.11
    0.5%
    46
    +0.11
    0.5%
    268
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    267
    +0.11
    0.03
    114
    +0.11
    0.02
    31
    +0.10
    0.02
    Negative Logits
     Decay
    -0.68
     Slam
    -0.61
    bats
    -0.60
     Steps
    -0.59
    bolt
    -0.59
     Trooper
    -0.57
     Gorge
    -0.57
     shaming
    -0.57
    liner
    -0.57
    fulness
    -0.56
    POSITIVE LOGITS
    orporated
    0.84
    reditary
    0.80
    rador
    0.77
    aughters
    0.76
    éĸ
    0.74
    ridor
    0.74
    ãĥīãĥ©ãĤ´ãĥ³
    0.73
    orpor
    0.73
    orea
    0.72
    ãĥ¼ãĥ«
    0.71
    Activations Density 0.151%

    No Known Activations