© 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
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 24576
    5. 8-RES_FS24576-JB
    6. 24135
    Prev
    Next
    INDEX
    Explanations

    occurrences of the word "cont" and its variations, suggesting it is focused on continuity or connection in text

    oai_token-act-pair · gpt-4o-miniTriggered 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
    167
    +0.15
    0.7%
    346
    +0.11
    0.5%
    232
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    232
    +0.15
    0.02
    167
    +0.11
    0.01
    137
    +0.10
    0.01
    Negative Logits
    EStream
    -0.79
    lihood
    -0.75
    manship
    -0.74
     scratch
    -0.69
    STER
    -0.67
    ãĤ¬
    -0.66
    OPLE
    -0.65
    terday
    -0.64
    ħĭ
    -0.63
     Lans
    -0.63
    POSITIVE LOGITS
    ributed
    1.44
    roversial
    1.39
    ribution
    1.35
    rovers
    1.33
    ribute
    1.27
    ainers
    1.26
    ainment
    1.24
    rast
    1.24
    emporary
    1.23
    acts
    1.20
    Activations Density 0.020%

    No Known Activations