© 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 EvalsExportsinterp-engineNEWAPI 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. 17661
    Prev
    Next
    INDEX
    Explanations

    specific instances of the word "which"

    oai_token-act-pair · gpt-3.5-turbo

    phrases that include the word "which."

    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
    let
    -0.72
    athi
    -0.70
    ben
    -0.67
    Behind
    -0.66
    belt
    -0.65
    Rog
    -0.63
    apt
    -0.62
    rolet
    -0.62
    bas
    -0.62
    politics
    -0.62
    POSITIVE LOGITS
    soever
    0.97
    xual
    0.90
    ãĥ¯ãĥ³
    0.71
    ĸļ
    0.69
     contrasts
    0.66
    earch
    0.64
    akespeare
    0.64
     consists
    0.64
     consisted
    0.63
    ilibrium
    0.62
    Activations Density 0.119%

    No Known Activations