© 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 EvalsExportsAPI 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. 28820
    Prev
    Next
    INDEX
    Explanations

    negative or critical sentiments related to individuals or entities

    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
    764
    +0.11
    0.5%
    304
    +0.10
    0.5%
    224
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    314
    +0.11
    0.00
    503
    +0.10
    0.04
    232
    +0.10
    0.08
    Negative Logits
     extinct
    -0.77
    icide
    -0.74
     raping
    -0.74
     surplus
    -0.73
     thieves
    -0.72
     unlucky
    -0.71
     robbers
    -0.70
     Solitaire
    -0.69
     predators
    -0.69
     melting
    -0.68
    POSITIVE LOGITS
    Comment
    1.01
    argument
    0.99
     Cosponsors
    0.94
     emph
    0.94
     Saying
    0.93
     remarks
    0.92
     rhet
    0.91
     succinct
    0.90
    reply
    0.90
    okingly
    0.90
    Activations Density 1.000%

    No Known Activations