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

    Neuronpedia

    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. 22436
    Prev
    Next
    INDEX
    Explanations

    exclamations and emphasis words

    oai_token-act-pair · gpt-3.5-turbo

    emphasized expressions of frustration or emphasis, particularly with the word "damn."

    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
    NetMessage
    -1.00
    ramid
    -0.75
    CRE
    -0.72
    anwhile
    -0.69
     membr
    -0.68
    KER
    -0.67
    cn
    -0.67
    KY
    -0.66
     Journals
    -0.65
    Interstitial
    -0.65
    POSITIVE LOGITS
     darn
    0.94
     damn
    0.90
    holes
    0.80
    ibly
    0.79
     damned
    0.79
    wit
    0.76
    selves
    0.75
    nuts
    0.74
    ighty
    0.73
    atio
    0.72
    Activations Density 0.010%

    No Known Activations