© 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 12288
    9. 8-RES_FS12288-JB
    11. 1482
    Prev
    Next
    INDEX
    Explanations

    phrases describing the use of something for a particular purpose for a specific group or cause

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    81
    +0.12
    0.5%
    440
    +0.12
    0.5%
    665
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    665
    +0.12
    0.05
    588
    +0.12
    0.04
    121
    +0.11
    0.04
    Negative Logits
     Lutheran
    -0.77
    Os
    -0.75
    Ox
    -0.73
    bey
    -0.73
     Aware
    -0.72
     Indigo
    -0.72
    76561
    -0.68
    opter
    -0.67
     Niet
    -0.65
    umar
    -0.65
    POSITIVE LOGITS
     blackmail
    0.96
     conjunction
    0.90
     pretext
    0.89
     marketing
    0.88
     leverage
    0.85
     purposes
    0.83
     coping
    0.83
     proxy
    0.80
     lieu
    0.80
     constructing
    0.79
    Activations Density 0.334%

    No Known Activations