Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlog/PodcastSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlog/PodcastGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    3. GPT2-Small
    4. Residual Stream
    5. 0-RES-JB
    6. 3270
    Prev
    Next
    INDEX
    Explanations

    words related to feet or shoes, especially focusing on the heel

    oai_token-act-pair · gpt-3.5-turbo

    references to "heel" and "parity" in various contexts

    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
     Corpus
    -0.97
    bris
    -0.82
     Wong
    -0.73
    ptives
    -0.72
    aris
    -0.71
    mers
    -0.69
     Explos
    -0.69
    maps
    -0.69
    Pir
    -0.68
     Schl
    -0.67
    POSITIVE LOGITS
     heel
    3.52
     redeem
    1.42
     redeemed
    1.36
     redemption
    1.28
     parity
    1.13
     feather
    0.96
     RAW
    0.96
    NBA
    0.92
    lace
    0.91
    IELD
    0.90
    Activations Density 0.059%

    No Known Activations