© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 49152
    9. 8-RES_FS49152-JB
    11. 42889
    Prev
    Next
    INDEX
    Explanations

    advancements in scientific research and technological innovations

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    480
    +0.26
    1.2%
    481
    +0.18
    0.8%
    447
    +0.14
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    647
    +0.26
    0.04
    642
    +0.18
    0.03
    106
    +0.14
    0.04
    Negative Logits
    é»Ĵ
    -0.77
    estate
    -0.72
    idays
    -0.72
     Franch
    -0.69
    fell
    -0.69
    jobs
    -0.68
    Street
    -0.68
     Situation
    -0.68
     Restaur
    -0.68
    76561
    -0.67
    POSITIVE LOGITS
     graphene
    1.29
     nanop
    1.20
     molecular
    1.19
     pept
    1.10
     chemically
    1.08
     nan
    1.06
     computational
    1.06
     enzyme
    1.06
     biom
    1.02
     genome
    1.02
    Activations Density 0.399%

    No Known Activations