© 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. 8-RES-JB
    11. 356
    Prev
    Next
    INDEX
    Explanations

    mentions of the Sundance Film Festival

    oai_token-act-pair · gpt-3.5-turbo

    references to the Sundance Film Festival

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    itably
    -0.74
     posit
    -0.69
     hemor
    -0.69
    course
    -0.64
    lectic
    -0.64
    HEAD
    -0.61
    odynam
    -0.61
    ACP
    -0.60
     clipboard
    -0.60
     Basketball
    -0.60
    POSITIVE LOGITS
    ered
    1.01
    inguished
    0.93
    ance
    0.87
    rained
    0.87
    Sund
    0.85
    sv
    0.82
    REAM
    0.81
     Sund
    0.80
    reth
    0.79
    ry
    0.78
    Activations Density 0.037%

    No Known Activations