© 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. 9819
    Prev
    Next
    INDEX
    Explanations

    proper nouns, such as names and places

    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
    87
    +0.43
    2.3%
    481
    +0.20
    1.1%
    519
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    87
    +0.43
    -0.01
    461
    +0.20
    0.06
    519
    +0.11
    0.06
    Negative Logits
     Morty
    -0.51
    OTA
    -0.50
     Augusta
    -0.50
     Anchorage
    -0.50
     Marlins
    -0.49
     Cassandra
    -0.48
     Homer
    -0.48
    Bitcoin
    -0.47
     Grassley
    -0.47
     âĢİ
    -0.47
    POSITIVE LOGITS
    fu
    0.55
    EStreamFrame
    0.53
     reps
    0.51
    ilant
    0.51
     initiation
    0.48
     HQ
    0.47
    achu
    0.47
    letes
    0.47
    pedia
    0.46
    naires
    0.46
    Activations Density 1.135%

    No Known Activations