© 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. 3-RES-JB
    11. 24518
    Prev
    Next
    INDEX
    Explanations

    actors' names

    oai_token-act-pair · gpt-3.5-turbo

    names of individuals or characters in a text

    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
    natureconservancy
    -0.92
    ources
    -0.78
    chwitz
    -0.78
    conom
    -0.75
     footing
    -0.69
     guiActiveUn
    -0.65
    andowski
    -0.65
     tremend
    -0.64
    ĻĤ
    -0.63
    urst
    -0.62
    POSITIVE LOGITS
     etc
    1.43
     Jr
    0.86
     respectively
    0.85
    etc
    0.84
     Guan
    0.73
     Sr
    0.73
     et
    0.73
     Yam
    0.70
     Sor
    0.67
     Metatron
    0.65
    Activations Density 0.468%

    No Known Activations