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. 373
    Prev
    Next
    INDEX
    Explanations

    sentences related to medical research and health conditions

    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
    430
    +0.12
    0.5%
    333
    +0.10
    0.5%
    84
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    426
    +0.12
    0.09
    430
    +0.10
    0.12
    546
    +0.10
    0.09
    Negative Logits
    etheless
    -0.91
    £ı
    -0.72
    NetMessage
    -0.66
    anuts
    -0.64
    osite
    -0.64
    ¬¼
    -0.62
    ĵĺ
    -0.62
    successfully
    -0.61
    enfranch
    -0.61
    »Ĵ
    -0.60
    POSITIVE LOGITS
     said
    1.33
    said
    1.29
     says
    1.23
     wrote
    1.14
     reads
    1.10
     commented
    0.99
     explained
    0.98
     writes
    0.96
     exclaimed
    0.96
     remarked
    0.95
    Activations Density 2.246%

    No Known Activations