© 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 12288
    9. 8-RES_FS12288-JB
    11. 453
    Prev
    Next
    INDEX
    Explanations

    statements related to physical health and recovery, particularly post-surgery and well-being updates

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    4
    +0.11
    0.5%
    448
    +0.10
    0.5%
    610
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    422
    +0.11
    0.04
    4
    +0.10
    0.06
    295
    +0.10
    0.04
    Negative Logits
     parody
    -0.80
     Insp
    -0.76
     indirectly
    -0.73
     ridicule
    -0.73
     scorn
    -0.71
     irony
    -0.70
     veto
    -0.69
     caricature
    -0.69
     poets
    -0.69
     monopol
    -0.69
    POSITIVE LOGITS
     stabilized
    1.22
     healed
    1.08
     functioning
    1.04
    stable
    1.02
     refreshed
    1.01
     stable
    1.01
     calmed
    1.01
     satisfactory
    0.97
     resumed
    0.97
    Ready
    0.95
    Activations Density 0.954%

    No Known Activations