© 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. 45813
    Prev
    Next
    INDEX
    Explanations

    phrases that indicate sourcing or referencing information

    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
    447
    +0.11
    0.5%
    88
    +0.10
    0.5%
    411
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    88
    +0.11
    0.02
    589
    +0.10
    0.01
    411
    +0.10
    0.01
    Negative Logits
    ANS
    -0.92
    accompan
    -0.86
    mire
    -0.84
    ans
    -0.80
    window
    -0.78
    dayName
    -0.75
    enth
    -0.74
     Cosponsors
    -0.74
    ourse
    -0.74
    anu
    -0.73
    POSITIVE LOGITS
     occasional
    1.12
     maybe
    0.86
     those
    0.83
     sporadic
    0.80
     inconvenience
    0.78
     perhaps
    0.75
     occasionally
    0.74
     anecdotal
    0.73
     its
    0.73
     superficial
    0.72
    Activations Density 0.020%

    No Known Activations