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. 21909
    Prev
    Next
    INDEX
    Explanations

    legal and governmental terms in a political context

    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
    671
    +0.10
    0.5%
    447
    +0.10
    0.5%
    72
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    665
    +0.10
    0.06
    608
    +0.10
    0.06
    61
    +0.10
    0.04
    Negative Logits
    interrupted
    -0.86
    ocard
    -0.69
    rolled
    -0.67
    gran
    -0.66
    spr
    -0.66
    Rating
    -0.65
    restricted
    -0.62
     Royale
    -0.62
    omet
    -0.62
    loo
    -0.60
    POSITIVE LOGITS
     substantive
    0.99
     acknowledge
    0.98
     adequately
    0.95
     whatsoever
    0.89
     nuance
    0.89
     acknowledging
    0.87
     nuanced
    0.87
     acknowledgement
    0.84
     facts
    0.84
     mention
    0.81
    Activations Density 0.478%

    No Known Activations