© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    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. 8-RES_POST_32K-OAI
    11. 5598
    Prev
    Next
    INDEX
    Explanations

    references to law enforcement and legal processes

    oai_token-act-pair · gpt-4o-miniTriggered 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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    373
    +0.16
    0.8%
    179
    +0.12
    0.5%
    424
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    146
    +0.16
    0.10
    288
    +0.12
    0.11
    224
    +0.11
    0.08
    Negative Logits
     bets
    -0.77
     smiles
    -0.75
     chants
    -0.74
     contrasts
    -0.71
     rows
    -0.69
     stares
    -0.68
     hugs
    -0.68
     sleeps
    -0.68
     youngsters
    -0.67
     stroll
    -0.65
    POSITIVE LOGITS
    mbuds
    0.88
    mbudsman
    0.84
     requesting
    0.82
     hotline
    0.81
     ASAP
    0.76
    istries
    0.76
     Registrar
    0.72
     Office
    0.70
     Directorate
    0.69
     Registry
    0.65
    Activations Density 1.289%

    No Known Activations