© 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. 4-RES-JB
    11. 1924
    Prev
    Next
    INDEX
    Explanations

    phrases related to filling in or completing information

    oai_token-act-pair · gpt-3.5-turbo

    instances of the phrase "fill in."

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

    No Comments

    Negative Logits
    VA
    -0.61
    ="/
    -0.58
    hetical
    -0.56
     TY
    -0.54
    href
    -0.52
    NAME
    -0.51
    lette
    -0.50
    TY
    -0.49
    DEN
    -0.49
     rating
    -0.49
    POSITIVE LOGITS
    umbn
    0.68
    vironment
    0.67
    othal
    0.67
    obyl
    0.62
    ulate
    0.61
    anship
    0.61
    senal
    0.60
    clusions
    0.59
    opal
    0.59
    ertodd
    0.59
    Activations Density 0.264%

    No Known Activations