© 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. 1-RES-JB
    11. 11621
    Prev
    Next
    INDEX
    Explanations

    names related to individuals involved in political news

    oai_token-act-pair · gpt-3.5-turbo

    mentions of Jared Kushner

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 1-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    1
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    cknowled
    -0.76
    Ķ
    -0.74
    ¡
    -0.72
    raged
    -0.71
     constitu
    -0.69
     mindless
    -0.65
    ¢
    -0.63
    ++++++++++++++++
    -0.63
    fight
    -0.61
    ľ
    -0.61
    POSITIVE LOGITS
     Kushner
    1.22
    doms
    0.93
    arov
    0.84
    gren
    0.82
    abba
    0.81
    lings
    0.81
    ules
    0.77
     Horowitz
    0.76
    icter
    0.75
    pai
    0.74
    Activations Density 0.005%

    No Known Activations