© 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. 10-RES-JB
    11. 20348
    Prev
    Next
    INDEX
    Explanations

    the number of times specific events or achievements occur, with a particular focus on sports successes

    oai_token-act-pair · gpt-3.5-turbo

    quantities and numerical achievements in sports or competitions

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 10-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    10
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    amaru
    -0.76
    fff
    -0.76
    Redditor
    -0.74
    FFER
    -0.73
    zona
    -0.71
    UGE
    -0.68
    rade
    -0.63
    rish
    -0.63
    atan
    -0.61
    abwe
    -0.61
    POSITIVE LOGITS
     consecutive
    1.17
     successive
    1.14
     separate
    1.01
     occasions
    0.99
     different
    0.95
     previous
    0.93
     other
    0.91
     times
    0.91
     straight
    0.91
     major
    0.90
    Activations Density 0.172%

    No Known Activations