© 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. 10242
    Prev
    Next
    INDEX
    Explanations

    scenarios related to personal experiences of abuse and criminal activities

    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
    725
    +0.11
    0.5%
    76
    +0.10
    0.4%
    371
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    710
    +0.11
    0.05
    54
    +0.10
    0.05
    554
    +0.10
    -0.01
    Negative Logits
    arations
    -0.81
    DNA
    -0.73
     extinct
    -0.69
     feasibility
    -0.69
     Components
    -0.67
    requisites
    -0.66
    inally
    -0.66
    illusion
    -0.65
     Corinth
    -0.65
    enfranch
    -0.65
    POSITIVE LOGITS
     gatherings
    1.10
     dinners
    1.04
     nightly
    1.04
     whenever
    1.03
     evenings
    1.01
     mornings
    1.01
     periodically
    0.97
     outings
    0.97
     specials
    0.96
     meetings
    0.96
    Activations Density 0.997%

    No Known Activations