© 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. 4-RES-JB
    11. 15455
    Prev
    Next
    INDEX
    Explanations

    phrases indicating completion or accomplishment

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "done."

    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
    lights
    -0.77
    wit
    -0.73
    ussen
    -0.73
    ulates
    -0.71
    olson
    -0.71
    uyomi
    -0.69
    uru
    -0.68
    liner
    -0.67
    uran
    -0.67
    pter
    -0.66
    POSITIVE LOGITS
    pez
    0.96
     away
    0.71
     nothing
    0.68
     outreach
    0.65
    ggie
    0.65
     brisk
    0.65
     prepar
    0.64
    omsday
    0.64
     toget
    0.64
    berman
    0.64
    Activations Density 0.026%

    No Known Activations