© 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 24576
    9. 8-RES_FS24576-JB
    11. 7279
    Prev
    Next
    INDEX
    Explanations

    phrases related to changes or modifications

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    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
    608
    +0.14
    0.6%
    139
    +0.09
    0.4%
    393
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    608
    +0.14
    0.03
    557
    +0.09
    0.02
    528
    +0.09
    0.02
    Negative Logits
    horn
    -0.82
    quickShipAvailable
    -0.75
     Cosponsors
    -0.64
    lehem
    -0.64
    MpServer
    -0.63
    thing
    -0.61
    Got
    -0.61
    CI
    -0.61
    acet
    -0.61
    wyn
    -0.59
    POSITIVE LOGITS
     accommodate
    1.11
     appease
    1.10
     ensure
    1.08
     enhance
    0.97
     reflect
    0.96
     clarify
    0.95
     conserve
    0.95
     improve
    0.93
     avoid
    0.93
     minimize
    0.92
    Activations Density 0.078%

    No Known Activations