© 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. 42913
    Prev
    Next
    INDEX
    Explanations

    instances of the word "new" in various contexts

    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
    443
    +0.13
    0.6%
    503
    +0.11
    0.5%
    132
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    443
    +0.13
    0.04
    339
    +0.11
    0.04
    489
    +0.11
    0.04
    Negative Logits
    ouse
    -0.77
    ...]
    -0.71
    acebook
    -0.69
    âĢ¦]
    -0.68
     Bastard
    -0.68
    ffield
    -0.67
    ohm
    -0.66
    ;;;;
    -0.66
    Zip
    -0.63
    osaurus
    -0.63
    POSITIVE LOGITS
     arrivals
    1.15
     revelations
    0.98
    bie
    0.96
    bies
    0.93
     entrants
    0.89
     hires
    0.86
     testament
    0.84
     developments
    0.84
     impetus
    0.82
     additions
    0.81
    Activations Density 0.070%

    No Known Activations