© 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. 3-RES-JB
    11. 17877
    Prev
    Next
    INDEX
    Explanations

    references to different levels and aspects of government

    oai_token-act-pair · gpt-3.5-turbo

    references to local government and media

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 3-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    3
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    yss
    -0.74
    bye
    -0.72
     personally
    -0.67
    idious
    -0.66
    tein
    -0.66
    ilitarian
    -0.64
     Valhalla
    -0.62
    ullivan
    -0.62
     Moonlight
    -0.61
     Wool
    -0.61
    POSITIVE LOGITS
    abulary
    0.71
    SPA
    0.71
    pers
    0.67
     mosqu
    0.66
     manag
    0.65
     affairs
    0.65
     regist
    0.65
     districts
    0.64
     veter
    0.64
     units
    0.63
    Activations Density 0.303%

    No Known Activations