© 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. 733
    Prev
    Next
    INDEX
    Explanations

    phrases related to evaluation or comparison

    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
    68
    +0.12
    0.5%
    254
    +0.10
    0.4%
    592
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    14
    +0.12
    0.04
    608
    +0.10
    0.06
    380
    +0.10
    0.03
    Negative Logits
    cation
    -0.93
    icism
    -0.92
    iliation
    -0.89
    terness
    -0.83
    oire
    -0.81
    culosis
    -0.81
    leness
    -0.80
    rency
    -0.79
    rower
    -0.79
    obook
    -0.78
    POSITIVE LOGITS
     constants
    1.05
     moments
    1.05
     truths
    1.00
     indications
    0.99
    types
    0.98
     reminders
    0.96
     exceptions
    0.95
     examples
    0.93
     occasions
    0.92
     embodiments
    0.91
    Activations Density 0.387%

    No Known Activations