© 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. 0-RES-JB
    11. 5012
    Prev
    Next
    INDEX
    Explanations

    phrases related to being complete or whole

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "full."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 0-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    0
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    abuse
    -0.70
     relates
    -0.67
    inois
    -0.66
    timer
    -0.65
    trap
    -0.64
     Guys
    -0.63
    cliffe
    -0.63
    atcher
    -0.62
    agine
    -0.62
    bourg
    -0.62
    POSITIVE LOGITS
     full
    3.51
    full
    2.39
     FULL
    2.30
    Full
    2.16
     Full
    2.10
     fullest
    2.01
     fuller
    1.92
     complete
    1.69
     fully
    1.41
     entirety
    1.30
    Activations Density 0.025%

    No Known Activations