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 3072
    5. 8-RES_FS3072-JB
    6. 983
    Prev
    Next
    INDEX
    Explanations

    programming concepts and methods, particularly related to working with C++ and symbols

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs3072-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    3,072
    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
    255
    +0.19
    0.9%
    326
    +0.14
    0.6%
    447
    +0.12
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    255
    +0.19
    0.13
    326
    +0.14
    0.12
    678
    +0.12
    0.08
    Negative Logits
    waters
    -0.76
    isine
    -0.74
    ussen
    -0.73
    smoking
    -0.73
    odcast
    -0.71
    fare
    -0.71
    brand
    -0.71
    study
    -0.71
     kitchens
    -0.70
    gaming
    -0.70
    POSITIVE LOGITS
     parameter
    1.40
     initialization
    1.37
     recursive
    1.34
     namespace
    1.31
     dynamically
    1.31
     syntax
    1.28
     JSON
    1.28
     boolean
    1.25
     integers
    1.25
     nested
    1.24
    Activations Density 1.879%

    No Known Activations