© 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 3072
    9. 8-RES_FS3072-JB
    11. 2987
    Prev
    Next
    INDEX
    Explanations

    references to dark or mysterious themes

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs3072-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    3,072
    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
    578
    +0.10
    0.4%
    288
    +0.08
    0.4%
    386
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.10
    0.08
    326
    +0.08
    0.06
    99
    +0.08
    0.04
    Negative Logits
    kson
    -0.77
    berman
    -0.73
    ciation
    -0.70
    utic
    -0.70
    oples
    -0.69
    Virtual
    -0.68
    agine
    -0.67
    UTERS
    -0.67
    aii
    -0.65
    onent
    -0.65
    POSITIVE LOGITS
    hound
    1.18
    horse
    1.00
    bolt
    0.86
    mere
    0.86
    croft
    0.83
    dark
    0.82
    moon
    0.81
     lord
    0.81
    urnal
    0.81
    crow
    0.80
    Activations Density 0.743%

    No Known Activations