© 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. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    5. GPT2-Small
    7. Residual Stream
    9. 8-RES-JB
    11. 11357
    Prev
    Next
    INDEX
    Explanations

    mentions related to the country Denmark

    oai_token-act-pair · gpt-3.5-turbo

    mentions of Denmark and its related entities

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.hook_resid_pre
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    plain
    -0.80
     Mayweather
    -0.71
    helle
    -0.71
    inventory
    -0.71
    <<
    -0.69
    reads
    -0.68
    DonaldTrump
    -0.67
    onse
    -0.66
    acco
    -0.65
    pite
    -0.65
    POSITIVE LOGITS
    borg
    1.17
     Bok
    0.95
     Helsinki
    0.90
     Finland
    0.89
     Scandinavian
    0.87
    å
    0.86
     Andersen
    0.85
     Danish
    0.85
     Ragnar
    0.85
     Denmark
    0.84
    Activations Density 0.051%

    No Known Activations