© 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. 28817
    Prev
    Next
    INDEX
    Explanations

    topics related to crimes and legal violations

    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
    628
    +0.14
    0.6%
    280
    +0.10
    0.5%
    419
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    314
    +0.14
    -0.01
    288
    +0.10
    0.11
    628
    +0.10
    0.06
    Negative Logits
     Glacier
    -0.73
    hof
    -0.73
     Metatron
    -0.71
     Fiorina
    -0.69
     srfAttach
    -0.68
     ellipt
    -0.66
     Pixie
    -0.66
     Valhalla
    -0.66
     toast
    -0.66
     Ivanka
    -0.66
    POSITIVE LOGITS
     crimes
    1.31
    crime
    1.29
     offences
    1.28
     offenses
    1.27
     crime
    1.26
     Crimes
    1.23
     offence
    1.20
     Crime
    1.18
     offenders
    1.17
    icide
    1.17
    Activations Density 1.283%

    No Known Activations