© 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
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 49152
    9. 8-RES_FS49152-JB
    11. 37397
    Prev
    Next
    INDEX
    Explanations

    the word "you" in various contexts

    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
    373
    +0.13
    0.6%
    747
    +0.08
    0.4%
    240
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    747
    +0.13
    0.01
    131
    +0.08
    0.01
    604
    +0.08
    0.01
    Negative Logits
     Lau
    -0.59
     Pratt
    -0.56
     Feld
    -0.55
    ignty
    -0.55
     Wonderland
    -0.54
     Vengeance
    -0.53
     Ratt
    -0.53
     Mig
    -0.53
    tsky
    -0.53
    escription
    -0.52
    POSITIVE LOGITS
    ा
    0.76
     sir
    0.67
    interstitial
    0.66
    RS
    0.64
    mbuds
    0.63
    à¥
    0.63
    tub
    0.63
    à¤
    0.62
    ãĥ¤
    0.60
     kindly
    0.60
    Activations Density 0.006%

    No Known Activations