© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 49152
    5. 8-RES_FS49152-JB
    6. 26583
    Prev
    Next
    INDEX
    Explanations

    instances of the word "said" and its variations, indicating reported speech or quotations

    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
    447
    +0.13
    0.6%
    421
    +0.10
    0.5%
    378
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    142
    +0.13
    0.05
    590
    +0.10
    0.04
    421
    +0.10
    0.04
    Negative Logits
    Pont
    -0.79
    tumblr
    -0.77
    «
    -0.75
    à¦
    -0.74
    ε
    -0.74
    ãĥĬ
    -0.73
    Advertisements
    -0.73
    Lens
    -0.72
    ãĥİ
    -0.72
    Ò
    -0.71
    POSITIVE LOGITS
     anecd
    0.87
     there
    0.69
     bluntly
    0.67
     he
    0.67
     although
    0.66
     Tuesday
    0.66
     Thursday
    0.65
     Hancock
    0.64
     Wednesday
    0.64
    heny
    0.63
    Activations Density 0.124%

    No Known Activations