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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlog/PodcastSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlog/PodcastGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 49152
    5. 8-RES_FS49152-JB
    6. 41347
    Prev
    Next
    INDEX
    Explanations

    instances of the word "a" across the text

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    36
    +0.22
    1.0%
    393
    +0.14
    0.6%
    362
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    36
    +0.22
    0.10
    159
    +0.14
    0.04
    362
    +0.12
    0.05
    Negative Logits
    aura
    -0.77
    alties
    -0.73
    lements
    -0.71
    isson
    -0.69
    upon
    -0.68
    olicy
    -0.68
    Links
    -0.68
    alion
    -0.68
    alias
    -0.68
    orders
    -0.67
    POSITIVE LOGITS
     beginner
    1.01
     novice
    0.95
     guy
    0.92
     politician
    0.91
     newborn
    0.89
     typical
    0.89
     woman
    0.88
     fledgling
    0.86
     nutshell
    0.86
     lot
    0.84
    Activations Density 0.168%

    No Known Activations