© 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. 2-RES-JB
    11. 11488
    Prev
    Next
    INDEX
    Explanations

    mentions of specific geographic locations, particularly Salt Lake City and Prague

    oai_token-act-pair · gpt-3.5-turbo

    references to the Salt Lake City area

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 2-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    2
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    merce
    -1.16
    mercial
    -0.94
    enegger
    -0.85
    ufact
    -0.81
     therap
    -0.76
    alez
    -0.75
    orders
    -0.72
     arrang
    -0.71
    schild
    -0.67
    GROUP
    -0.65
    POSITIVE LOGITS
     Salt
    1.11
    Salt
    0.98
    zman
    0.94
    water
    0.88
    itude
    0.88
     Lake
    0.86
    illo
    0.82
    ash
    0.80
    bite
    0.77
     baths
    0.75
    Activations Density 0.003%

    No Known Activations