© 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. 10-RES_POST_32K-OAI
    11. 31982
    Prev
    Next
    INDEX
    Explanations

    occurrences of the word "in" and variations related to legal or formal contexts

    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
    756
    +0.12
    0.6%
    235
    +0.10
    0.5%
    22
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    123
    +0.12
    0.14
    22
    +0.10
    0.12
    183
    +0.09
    0.09
    Negative Logits
     unemploy
    -0.76
     è£ıè
    -0.69
     Conserv
    -0.67
     Agric
    -0.64
     corpus
    -0.63
     Rebell
    -0.63
     charism
    -0.63
     toget
    -0.63
     Cth
    -0.63
     INST
    -0.62
    POSITIVE LOGITS
     remarks
    1.09
    aug
    0.91
     comments
    0.90
     announcing
    0.89
     Wednesday
    0.88
     Thursday
    0.88
     recent
    0.86
     tweets
    0.85
     Tuesday
    0.85
     interviews
    0.83
    Activations Density 0.685%

    No Known Activations