© 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. 1-RES-JB
    11. 2641
    Prev
    Next
    INDEX
    Explanations
    No Explanations Found
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 1-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    1
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    oler
    -0.77
    arta
    -0.67
    HI
    -0.66
     Certificate
    -0.66
    ELD
    -0.65
     guard
    -0.65
    cone
    -0.61
     Tend
    -0.61
    é¾įå¥ij士
    -0.59
    eller
    -0.59
    POSITIVE LOGITS
    reddits
    0.64
    ı
    0.64
    iq
    0.63
    iov
    0.63
    vell
    0.63
     skins
    0.62
    yrights
    0.62
     subreddit
    0.61
    arah
    0.61
    =/
    0.61
    Activations Density 0.000%

    No Known Activations

    This feature has no known activations.