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. 23311
    Prev
    Next
    INDEX
    Explanations

    mentions of exploitation or illegal schemes characterized by the term "racket."

    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
    587
    +0.14
    0.6%
    746
    +0.12
    0.5%
    128
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.14
    0.01
    326
    +0.12
    0.01
    596
    +0.11
    0.01
    Negative Logits
     AFB
    -0.76
    stock
    -0.71
    irtual
    -0.70
    shire
    -0.68
    itates
    -0.68
     Aram
    -0.67
    borgh
    -0.65
     experien
    -0.65
    ibility
    -0.64
    Mos
    -0.64
    POSITIVE LOGITS
    eering
    2.06
    eers
    1.98
    eer
    1.77
    ographed
    0.98
    folios
    0.84
    andum
    0.84
    eal
    0.82
    nels
    0.81
     racket
    0.81
    ALS
    0.79
    Activations Density 0.008%

    No Known Activations