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

    Neuronpedia

    Jacobian LensNEW
    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. 20311
    Prev
    Next
    INDEX
    Explanations

    phrases related to a midpoint or halfway point

    oai_token-act-pair · gpt-3.5-turbo

    phrases related to measurements or fractions of time and distance, particularly those involving "halfway."

    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
     Garland
    -0.66
     Citation
    -0.65
    ãĥ©
    -0.65
     Tags
    -0.64
    imaru
    -0.64
     Scrib
    -0.63
     QR
    -0.63
     Lowell
    -0.63
    PLA
    -0.63
    ¯¯¯¯
    -0.62
    POSITIVE LOGITS
    dozen
    0.92
     dozen
    0.91
    mort
    0.82
    tails
    0.77
    arb
    0.72
    ctic
    0.71
    imester
    0.71
    hops
    0.70
    ensive
    0.70
    nir
    0.69
    Activations Density 0.150%

    No Known Activations