© 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 EvalsExportsinterp-engineNEWAPI 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. 8653
    Prev
    Next
    INDEX
    Explanations

    phrases indicating emotional connections and personal relationships

    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
    75
    +0.10
    0.5%
    751
    +0.10
    0.4%
    23
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    6
    +0.10
    0.23
    640
    +0.10
    0.24
    430
    +0.09
    0.22
    Negative Logits
     sisters
    -0.96
     girls
    -0.89
     femin
    -0.83
    cca
    -0.82
     nuns
    -0.81
    girls
    -0.80
    maid
    -0.79
     women
    -0.78
     herself
    -0.77
     sister
    -0.76
    POSITIVE LOGITS
    Mr
    1.04
     Manziel
    0.97
     Mr
    0.94
     Rodrigo
    0.93
    His
    0.90
     Duterte
    0.88
    He
    0.83
    HAEL
    0.82
    uterte
    0.82
    iannopoulos
    0.80
    Activations Density 6.869%

    No Known Activations