© 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. 9753
    Prev
    Next
    INDEX
    Explanations

    legal hearings and courtroom proceedings

    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
    447
    +0.16
    0.7%
    195
    +0.12
    0.5%
    310
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    500
    +0.16
    0.03
    310
    +0.12
    0.03
    395
    +0.11
    0.02
    Negative Logits
    jac
    -0.73
    acea
    -0.72
    rave
    -0.68
    ienne
    -0.67
    ecd
    -0.66
    eros
    -0.65
    obs
    -0.65
    ijn
    -0.64
    oil
    -0.64
    spr
    -0.64
    POSITIVE LOGITS
     hearings
    1.27
     testimony
    1.07
     examiner
    1.02
     transcripts
    1.00
     Hearing
    0.98
     proceedings
    0.98
     convened
    0.95
     adjourn
    0.94
     hearing
    0.93
    imony
    0.92
    Activations Density 0.066%

    No Known Activations