© 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. 12475
    Prev
    Next
    INDEX
    Explanations

    mentions of two items, concepts, or entities occurring together or in relation to each other

    oai_token-act-pair · gpt-3.5-turbo

    phrases that include the word "and."

    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
    ³
    -0.80
    º
    -0.73
    ĨĴ
    -0.72
    odore
    -0.72
    Ĥİ
    -0.72
    iane
    -0.72
    į
    -0.71
    uel
    -0.71
    Ļ½
    -0.70
    ½
    -0.69
    POSITIVE LOGITS
     sexes
    0.91
     halves
    0.82
     sides
    0.77
     genders
    0.77
     equally
    0.72
     disciplines
    0.65
     pronounced
    0.63
    igham
    0.62
     amuse
    0.62
    ruciating
    0.61
    Activations Density 0.119%

    No Known Activations