© 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. 3-RES-JB
    11. 13038
    Prev
    Next
    INDEX
    Explanations

    references to a specific geographical location, Fort Detrick

    oai_token-act-pair · gpt-3.5-turbo

    references to "Det" and related terms, likely indicating a focus on details regarding "Detainers" in a legal or military context

    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
    xual
    -1.02
    manship
    -0.86
    nown
    -0.85
    ammy
    -0.81
    hetti
    -0.77
    drive
    -0.76
    lihood
    -0.75
     deaf
    -0.71
    externalActionCode
    -0.68
     tremend
    -0.68
    POSITIVE LOGITS
    ector
    1.21
    ainers
    0.97
    ention
    0.96
    ailed
    0.96
    rans
    0.95
    roit
    0.92
    agon
    0.90
    achment
    0.88
    ective
    0.87
    ected
    0.86
    Activations Density 0.013%

    No Known Activations