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. 12035
    Prev
    Next
    INDEX
    Explanations

    verb forms in various tenses, indicating actions or states

    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
    314
    +0.16
    0.8%
    55
    +0.13
    0.6%
    373
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    55
    +0.16
    0.07
    102
    +0.13
    0.07
    566
    +0.12
    0.04
    Negative Logits
     Vers
    -0.72
     SPD
    -0.69
     Notting
    -0.65
     Benn
    -0.64
     Dunk
    -0.64
     dismant
    -0.63
     predicate
    -0.62
     Hust
    -0.62
     millenn
    -0.61
     juven
    -0.59
    POSITIVE LOGITS
    tics
    0.91
    senal
    0.89
    Ĥİ
    0.89
     guiActiveUnfocused
    0.86
    rael
    0.80
    hyde
    0.79
    abella
    0.76
    ifles
    0.75
    tis
    0.74
    ws
    0.74
    Activations Density 0.306%

    No Known Activations