© 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. 3044
    Prev
    Next
    INDEX
    Explanations

    information related to construction projects and infrastructure developments

    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
    41
    +0.10
    0.4%
    325
    +0.10
    0.4%
    761
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    113
    +0.10
    0.07
    41
    +0.10
    0.07
    325
    +0.09
    0.08
    Negative Logits
     diapers
    -0.78
    papers
    -0.71
    onom
    -0.69
    appropriate
    -0.67
     dolls
    -0.67
     diaper
    -0.67
    ealous
    -0.67
    CEO
    -0.66
    confidence
    -0.66
    azz
    -0.65
    POSITIVE LOGITS
     Route
    1.47
    Route
    1.45
     Highway
    1.36
     Interstate
    1.36
     junction
    1.30
     east
    1.29
    outheast
    1.29
     highways
    1.28
     corridor
    1.23
     west
    1.20
    Activations Density 2.062%

    No Known Activations