© 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 49152
    9. 8-RES_FS49152-JB
    11. 37495
    Prev
    Next
    INDEX
    Explanations

    directional language and movement-related phrases

    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
    481
    +0.17
    0.8%
    373
    +0.11
    0.5%
    127
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    526
    +0.17
    0.00
    113
    +0.11
    0.04
    323
    +0.10
    0.04
    Negative Logits
     grouping
    -0.71
    saf
    -0.65
     referring
    -0.64
     appointing
    -0.63
    cia
    -0.61
     distinguishing
    -0.61
    uld
    -0.59
    onom
    -0.59
    amera
    -0.59
     sparing
    -0.58
    POSITIVE LOGITS
     cig
    0.73
    aband
    0.69
    iped
    0.69
     footsteps
    0.67
     pipes
    0.67
    tics
    0.65
     stages
    0.65
     graves
    0.64
     torches
    0.64
     passers
    0.64
    Activations Density 0.340%

    No Known Activations