© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 22539
    Prev
    Next
    INDEX
    Explanations

    words related to a specific sports team or action associated with ducks

    oai_token-act-pair · gpt-3.5-turbo

    references to certain sports teams, particularly hockey teams

    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
     stage
    -0.79
     condition
    -0.68
     pane
    -0.67
     Form
    -0.65
     sought
    -0.65
     Crom
    -0.63
     Rey
    -0.63
     later
    -0.62
     cycle
    -0.62
     Col
    -0.62
    POSITIVE LOGITS
    ucks
    4.53
    uck
    2.13
    UCK
    1.87
    ucking
    1.73
    ucked
    1.53
    ucker
    1.31
    ucky
    1.23
    licks
    1.16
    okin
    1.08
    ocks
    1.07
    Activations Density 0.014%

    No Known Activations