© 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. 20252
    Prev
    Next
    INDEX
    Explanations

    proper nouns and names of people or places associated with current events or news stories

    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
    393
    +0.15
    0.7%
    362
    +0.13
    0.6%
    745
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    362
    +0.15
    0.05
    385
    +0.13
    0.03
    393
    +0.12
    0.03
    Negative Logits
    tle
    -0.72
    oka
    -0.69
     suppose
    -0.66
    arget
    -0.64
    thood
    -0.64
    ich
    -0.64
    iru
    -0.63
    Percent
    -0.63
    zu
    -0.63
     namely
    -0.62
    POSITIVE LOGITS
     entirety
    1.12
     midst
    1.08
     veins
    1.06
     motions
    1.04
     ranks
    1.02
     backdoor
    1.02
     corridors
    1.01
     halls
    1.00
     process
    0.98
     cracks
    0.97
    Activations Density 0.099%

    No Known Activations