© 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 24576
    9. 8-RES_FS24576-JB
    11. 7368
    Prev
    Next
    INDEX
    Explanations

    phrases related to staying on the same path or staying committed

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    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
    430
    +0.14
    0.6%
    725
    +0.10
    0.5%
    352
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    422
    +0.14
    0.02
    567
    +0.10
    0.02
    298
    +0.10
    0.02
    Negative Logits
     descending
    -0.67
     Macintosh
    -0.60
     soph
    -0.59
     Peg
    -0.59
    geries
    -0.58
     dash
    -0.57
     Ferguson
    -0.57
     negotiators
    -0.57
     disregard
    -0.57
     regression
    -0.56
    POSITIVE LOGITS
    lishes
    0.89
    leep
    0.84
    eper
    0.78
    cation
    0.77
    til
    0.74
    hook
    0.72
    cool
    0.71
    inet
    0.65
    addons
    0.65
    stay
    0.64
    Activations Density 0.071%

    No Known Activations