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 24576
    5. 8-RES_FS24576-JB
    6. 23851
    Prev
    Next
    INDEX
    Explanations

    instances of the word "rather."

    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
    481
    +0.11
    0.5%
    52
    +0.11
    0.5%
    99
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    52
    +0.11
    0.02
    49
    +0.11
    0.02
    406
    +0.11
    0.02
    Negative Logits
    elsen
    -0.97
    iliate
    -0.84
    adium
    -0.82
    amba
    -0.80
    ividual
    -0.79
    eka
    -0.75
    ajo
    -0.75
    ULTS
    -0.75
    esville
    -0.74
    esi
    -0.74
    POSITIVE LOGITS
     amusing
    1.00
     inconvenient
    0.97
     unpleasant
    0.93
     pointless
    0.89
     rude
    0.88
     pathetic
    0.88
     nifty
    0.86
     unrem
    0.85
     tame
    0.85
     awkward
    0.84
    Activations Density 0.017%

    No Known Activations