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 · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 24576
    5. 8-RES_FS24576-JB
    6. 683
    Prev
    Next
    INDEX
    Explanations

    phrases related to legal and social issues related to reproduction and privacy, particularly focusing on fertility clinics, gay rights, and privacy

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    481
    +0.23
    1.1%
    447
    +0.11
    0.5%
    169
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.23
    0.08
    147
    +0.11
    0.04
    508
    +0.11
    0.04
    Negative Logits
    lihood
    -0.85
     TAMADRA
    -0.77
     Harden
    -0.67
     Presence
    -0.65
    iphate
    -0.62
    given
    -0.62
     Nebula
    -0.62
     Rampage
    -0.61
     Offensive
    -0.61
     Prelude
    -0.61
    POSITIVE LOGITS
     professionals
    1.29
     experts
    1.28
     enthusiasts
    1.24
     historians
    1.17
     specialists
    1.17
     campaigners
    1.13
     analysts
    1.13
     regulators
    1.13
     agencies
    1.13
     practitioners
    1.12
    Activations Density 0.369%

    No Known Activations