© 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. 27663
    Prev
    Next
    INDEX
    Explanations

    conversational elements and dialogue markers in text

    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
    56
    +0.11
    0.5%
    8
    +0.10
    0.5%
    588
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    107
    +0.11
    0.11
    283
    +0.10
    0.03
    8
    +0.10
    0.09
    Negative Logits
     distorted
    -0.92
     incorrectly
    -0.91
     inaccur
    -0.87
     suppressed
    -0.86
     errone
    -0.86
     deteriorated
    -0.83
     retracted
    -0.81
     weakened
    -0.81
     worsened
    -0.80
     erroneous
    -0.79
    POSITIVE LOGITS
    morrow
    1.01
     preferably
    1.00
    Tomorrow
    0.99
    Sure
    0.98
     Bye
    0.94
     promise
    0.93
     TBA
    0.91
     tomorrow
    0.86
    Cool
    0.86
     Promise
    0.85
    Activations Density 2.177%

    No Known Activations