Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 49152
    5. 8-RES_FS49152-JB
    6. 12115
    Prev
    Next
    INDEX
    Explanations

    instances of a specific symbol or character repeated in the text

    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
    23
    +0.10
    0.5%
    365
    +0.10
    0.5%
    103
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    23
    +0.10
    0.07
    352
    +0.10
    0.07
    103
    +0.09
    0.07
    Negative Logits
     scatter
    -0.82
    anwhile
    -0.78
    ettings
    -0.77
     afore
    -0.69
     rooting
    -0.68
    theless
    -0.68
     stomp
    -0.67
     shack
    -0.66
     scattering
    -0.64
     mileage
    -0.64
    POSITIVE LOGITS
    ¡
    1.23
    į
    1.19
    º
    1.16
    ²
    1.15
    ¥
    1.14
    ¹
    1.11
    ¬
    1.11
    Ĵ
    1.10
    Ń
    1.09
    «
    1.09
    Activations Density 0.105%

    No Known Activations