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

    Neuronpedia

    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 24576
    9. 8-RES_FS24576-JB
    11. 9829
    Prev
    Next
    INDEX
    Explanations

    prepositions and phrases indicating spatial relationships

    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
    289
    +0.12
    0.5%
    674
    +0.12
    0.5%
    459
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    289
    +0.12
    0.02
    53
    +0.12
    0.04
    469
    +0.11
    0.03
    Negative Logits
    acus
    -0.80
    fb
    -0.76
    idays
    -0.75
    abilities
    -0.72
    ivals
    -0.72
    deal
    -0.72
    isodes
    -0.68
    ilyn
    -0.68
    ologies
    -0.67
    çīĪ
    -0.66
    POSITIVE LOGITS
     molten
    0.96
     sorts
    0.92
     flats
    0.90
     rubble
    0.90
     sand
    0.89
     asphalt
    0.88
     glass
    0.88
     pixels
    0.86
     flesh
    0.85
     ice
    0.84
    Activations Density 0.097%

    No Known Activations