© 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. 4548
    Prev
    Next
    INDEX
    Explanations

    filler words and hesitations like "um" and "uh"

    oai_token-act-pair · gpt-3.5-turbo

    filler words or discourse markers in conversation

    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
    hips
    -0.95
    Kings
    -0.81
    BOOK
    -0.73
    ingham
    -0.71
     Dayton
    -0.67
    Constructed
    -0.66
    hire
    -0.65
    bedroom
    -0.65
    hani
    -0.64
    Born
    -0.64
    POSITIVE LOGITS
    ugh
    0.80
    pty
    0.80
     fu
    0.80
    ming
    0.76
     um
    0.74
    ilo
    0.74
    othe
    0.73
    bum
    0.72
    uh
    0.71
    essage
    0.71
    Activations Density 0.008%

    No Known Activations