© 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. 8-RES_POST_32K-OAI
    11. 7281
    Prev
    Next
    INDEX
    Explanations

    phrases that express commitment or dedication to a cause or action

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_post_32k-oai
    Configuration
    jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_8.pt
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.8.hook_resid_post
    Hook Layer
    8
    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
    74
    +0.10
    0.5%
    531
    +0.09
    0.4%
    150
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    608
    +0.10
    0.05
    665
    +0.09
    0.04
    385
    +0.09
    0.03
    Negative Logits
    DragonMagazine
    -1.21
    æĺ
    -0.89
    isen
    -0.77
    igmatic
    -0.77
    acements
    -0.76
    ults
    -0.75
    immer
    -0.71
    ratulations
    -0.70
    uman
    -0.69
    nette
    -0.69
    POSITIVE LOGITS
     preserving
    1.23
     ensuring
    1.06
     keeping
    1.03
     upholding
    1.02
     maintaining
    1.02
     protecting
    1.01
     secrecy
    0.99
     preventing
    0.98
     improving
    0.98
     helping
    0.96
    Activations Density 0.192%

    No Known Activations