© 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 EvalsExportsinterp-engineNEWAPI 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. 20779
    Prev
    Next
    INDEX
    Explanations

    phrases related to specific political figures or organizations, specifically "Ansar Dine" and "Ansar" with high activations, as well as other political terms like "playbook"

    oai_token-act-pair · gpt-3.5-turbo

    references to a specific entity or character named "Ans."

    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
     Canaver
    -0.70
     Durham
    -0.69
     Scotia
    -0.67
     Blasio
    -0.63
     GEAR
    -0.63
     Clyde
    -0.62
     Tycoon
    -0.62
     Clifford
    -0.61
     fluids
    -0.61
     CHR
    -0.61
    POSITIVE LOGITS
    wered
    1.41
     Ans
    1.23
    linger
    0.98
    ensibly
    0.92
    illus
    0.88
    anta
    0.87
    combe
    0.86
    unta
    0.85
    bach
    0.84
    entials
    0.84
    Activations Density 0.007%

    No Known Activations