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 768
    5. 8-RES_FS768-JB
    6. 119
    Prev
    Next
    INDEX
    Explanations

    mentions of legal cases and political events

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs768-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    768
    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
    +0.18
    0.8%
    401
    +0.12
    0.5%
    580
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.18
    0.24
    481
    +0.12
    0.21
    325
    +0.11
    0.16
    Negative Logits
    Aut
    -0.58
     Siber
    -0.56
    Ball
    -0.56
    param
    -0.54
     reconc
    -0.54
     ank
    -0.53
     Pil
    -0.53
    sers
    -0.52
    ICAL
    -0.52
    Nav
    -0.51
    POSITIVE LOGITS
     attest
    0.86
     ought
    0.83
     fame
    0.81
     outwe
    0.81
     outweigh
    0.81
     ensured
    0.78
     were
    0.77
     appeared
    0.76
     are
    0.76
     tended
    0.76
    Activations Density 6.260%

    No Known Activations