© 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. 3-RES-JB
    11. 23589
    Prev
    Next
    INDEX
    Explanations

    mentions of places where people reside or stay, such as residences, halls, or campuses

    oai_token-act-pair · gpt-3.5-turbo

    occurrences of the word "residence" in various contexts

    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
    umbs
    -0.74
    romeda
    -0.74
     Cosponsors
    -0.72
    reg
    -0.71
    displayText
    -0.71
    anson
    -0.68
    Cola
    -0.67
    aneously
    -0.65
    ammy
    -0.64
    CAST
    -0.63
    POSITIVE LOGITS
     halls
    0.96
     residence
    0.96
    maid
    0.88
    chool
    0.85
     Halls
    0.80
     hall
    0.75
     address
    0.74
    loo
    0.74
     dwelling
    0.73
    coming
    0.73
    Activations Density 0.007%

    No Known Activations