© 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 1536
    9. 8-RES_FS1536-JB
    11. 78
    Prev
    Next
    INDEX
    Explanations

    references to sports teams and related statistics

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs1536-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    1,536
    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
    139
    +0.14
    0.6%
    354
    +0.13
    0.6%
    481
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    354
    +0.14
    0.20
    139
    +0.13
    0.17
    697
    +0.12
    0.14
    Negative Logits
     Millennium
    -0.69
    Untitled
    -0.67
     Eleven
    -0.67
     Amen
    -0.65
    earch
    -0.63
     Upper
    -0.63
     Nin
    -0.62
    tones
    -0.62
     Innocent
    -0.62
    CAP
    -0.58
    POSITIVE LOGITS
    'll
    1.45
    've
    1.39
    're
    1.26
    'd
    1.25
     didn
    1.14
     certainly
    1.07
     shouldn
    1.05
    pherd
    1.04
     couldn
    1.03
     knew
    1.02
    Activations Density 1.988%

    No Known Activations