© 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. 824
    Prev
    Next
    INDEX
    Explanations

    detailed technical explanations or descriptions

    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
    302
    +0.14
    0.7%
    195
    +0.14
    0.6%
    186
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    302
    +0.14
    0.04
    195
    +0.14
    0.02
    455
    +0.10
    0.03
    Negative Logits
    corn
    -0.82
    enos
    -0.78
    avis
    -0.74
    ques
    -0.74
    aden
    -0.70
    soDeliveryDate
    -0.69
    haw
    -0.66
    Jar
    -0.66
    hops
    -0.65
    itto
    -0.64
    POSITIVE LOGITS
    bred
    1.08
     examination
    1.08
     descriptions
    1.01
     detail
    0.99
     exposition
    0.97
     description
    0.93
     interrogation
    0.90
     overview
    0.90
     detailing
    0.88
     explanation
    0.87
    Activations Density 0.122%

    No Known Activations