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 3072
    5. 8-RES_FS3072-JB
    6. 1811
    Prev
    Next
    INDEX
    Explanations

    phrases indicating exclusivity or singularity

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs3072-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    3,072
    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
    373
    +0.12
    0.5%
    756
    +0.12
    0.5%
    498
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    498
    +0.12
    0.08
    413
    +0.12
    0.06
    370
    +0.11
    0.05
    Negative Logits
    insula
    -0.73
    idon
    -0.73
    align
    -0.64
    comb
    -0.61
    ducers
    -0.61
    ahead
    -0.60
    hement
    -0.59
    phen
    -0.58
    finder
    -0.58
    dict
    -0.57
    POSITIVE LOGITS
     marginally
    1.05
     kidding
    0.85
    ices
    0.78
    ICES
    0.78
     lasted
    0.75
     incidentally
    0.74
     scratches
    0.74
     lasts
    0.73
     scratched
    0.73
     spor
    0.71
    Activations Density 0.370%

    No Known Activations