© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 0-RES-JB
    11. 18208
    Prev
    Next
    INDEX
    Explanations

    words related to the concept of "madness."

    oai_token-act-pair · gpt-3.5-turbo

    the repeated mention or emphasis of the word "Mad."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 0-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    0
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ctr
    -0.92
    tle
    -0.67
    ypes
    -0.67
     skin
    -0.66
    REF
    -0.66
    rf
    -0.64
    toc
    -0.64
    ebin
    -0.64
    EE
    -0.63
    comp
    -0.62
    POSITIVE LOGITS
     Mad
    3.62
    Mad
    2.92
     MAD
    1.80
    mad
    1.78
     mad
    1.69
     Madame
    1.34
     Madonna
    1.25
     Madd
    1.23
     Mud
    1.18
     Hom
    1.12
    Activations Density 0.024%

    No Known Activations