© 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. 26872
    Prev
    Next
    INDEX
    Explanations

    terms related to changes or comparisons in status or condition, particularly indicating new or existing elements

    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
    362
    +0.25
    1.2%
    408
    +0.14
    0.7%
    68
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    362
    +0.25
    0.11
    408
    +0.14
    0.09
    288
    +0.10
    0.09
    Negative Logits
    another
    -0.90
    ecause
    -0.87
     besides
    -0.86
     indeed
    -0.84
    formerly
    -0.82
    actionDate
    -0.81
    helm
    -0.81
     furthermore
    -0.81
    ntil
    -0.80
     therefore
    -0.79
    POSITIVE LOGITS
     ones
    1.06
     trio
    0.94
     amount
    0.93
     portion
    0.91
     confines
    0.89
     timeframe
    0.85
     brunt
    0.84
     contents
    0.84
     workings
    0.84
    bies
    0.79
    Activations Density 0.766%

    No Known Activations