© 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. 10-RES_POST_32K-OAI
    11. 29308
    Prev
    Next
    INDEX
    Explanations

    language related to moral judgment and ethical behavior

    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
    464
    +0.14
    0.7%
    762
    +0.13
    0.6%
    161
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    762
    +0.14
    0.07
    55
    +0.13
    0.06
    718
    +0.10
    0.05
    Negative Logits
    Tokens
    -0.87
    doesn
    -0.80
    itivity
    -0.80
    Steve
    -0.78
     GoPro
    -0.72
    Wow
    -0.71
    Nick
    -0.70
    anmar
    -0.69
     Nope
    -0.68
    Lots
    -0.68
    POSITIVE LOGITS
     acquainted
    1.10
    ,—
    1.08
     endeav
    1.08
     consec
    1.07
     emanc
    1.07
     furnished
    1.06
     prevailed
    1.00
     disposed
    1.00
     comprehend
    1.00
     apprehended
    0.98
    Activations Density 0.336%

    No Known Activations