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 1536
    5. 8-RES_FS1536-JB
    6. 926
    Prev
    Next
    INDEX
    Explanations

    the word "red" and related terms

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs1536-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    1,536
    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
    459
    +0.19
    0.9%
    12
    +0.10
    0.4%
    289
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    459
    +0.19
    0.09
    12
    +0.10
    0.07
    289
    +0.10
    0.03
    Negative Logits
    =-=-=-=-
    -0.73
    Ö¼
    -0.72
    =-=-
    -0.71
    ilities
    -0.70
    ILY
    -0.70
    ernel
    -0.69
    rolet
    -0.68
    agall
    -0.67
    uador
    -0.67
    gerald
    -0.63
    POSITIVE LOGITS
    irection
    1.05
    beard
    1.05
    efined
    1.04
    neck
    1.04
    berry
    1.03
    iscovered
    0.99
     velvet
    0.99
    eem
    0.98
    oubt
    0.98
    prints
    0.95
    Activations Density 1.109%

    No Known Activations