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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 8199
    Prev
    Next
    INDEX
    Explanations

    references to the natural disaster Hurricane Katrina

    oai_token-act-pair · gpt-3.5-turbo
    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
    enhagen
    -0.95
    omething
    -0.82
    initely
    -0.81
    reddits
    -0.80
    izoph
    -0.79
    emetery
    -0.79
    icient
    -0.78
    reads
    -0.78
    ön
    -0.77
    icult
    -0.75
    POSITIVE LOGITS
     Pact
    0.75
    âĸ¬
    0.74
     Cassidy
    0.74
     Hath
    0.74
     Samson
    0.72
     ç¥ŀ
    0.69
    ilee
    0.69
     compulsion
    0.65
    rawdownloadcloneembedreportprint
    0.64
     Gas
    0.64
    Activations Density 0.012%

    No Known Activations