© 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. 29059
    Prev
    Next
    INDEX
    Explanations

    phrases that emphasize the concept of "more" or suggest an increase in quantity or degree

    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
    190
    +0.11
    0.5%
    633
    +0.10
    0.4%
    647
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    190
    +0.11
    0.08
    159
    +0.10
    0.06
    102
    +0.09
    0.07
    Negative Logits
     wonders
    -0.85
     minds
    -0.69
    etimes
    -0.69
    great
    -0.67
     lousy
    -0.67
     Shares
    -0.67
     unworthy
    -0.67
     horrors
    -0.66
    utterstock
    -0.66
     evils
    -0.65
    POSITIVE LOGITS
    rounded
    0.92
     balanced
    0.86
     rounded
    0.80
     relaxed
    0.78
     matured
    0.76
     streamlined
    0.75
    agon
    0.73
    atin
    0.73
    atory
    0.72
     polished
    0.72
    Activations Density 0.500%

    No Known Activations