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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    3. GPT2-Small
    4. Residual Stream
    5. 11-RES-JB
    6. 10474
    Prev
    Next
    INDEX
    Explanations

    phrases related to different qualities or levels (high, middle, low, late) paired with a form of selection or ranking (prospects, end, mid)

    oai_token-act-pair · gpt-3.5-turbo

    phrases referring to player rankings and evaluations in sports

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 11-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    11
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     Thrones
    -0.82
     fame
    -0.70
     Cree
    -0.69
     voic
    -0.69
     Lords
    -0.67
     TED
    -0.65
     Boko
    -0.64
     Invisible
    -0.63
     Uganda
    -0.62
     Duchess
    -0.61
    POSITIVE LOGITS
    season
    1.10
    nine
    1.04
    league
    1.04
    division
    1.03
    six
    1.03
    seven
    1.03
    arching
    1.02
    half
    1.00
    arb
    1.00
    eligible
    0.99
    Activations Density 0.128%

    No Known Activations