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. 8494
    Prev
    Next
    INDEX
    Explanations

    terms related to legal and governmental departments and activities

    oai_token-act-pair · gpt-3.5-turboTriggered 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
    146
    +0.13
    0.6%
    288
    +0.12
    0.6%
    46
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    146
    +0.13
    0.05
    708
    +0.12
    0.03
    424
    +0.10
    0.03
    Negative Logits
    ãĥ¼ãĥĨ
    -0.84
    plet
    -0.79
    plets
    -0.74
    ãĥ¼ãĥ
    -0.73
    ãĥİ
    -0.72
    ãĤ¯
    -0.72
    Sov
    -0.71
    ãĤ¨ãĥ«
    -0.70
    jad
    -0.67
    ãĤ¤ãĥĪ
    -0.67
    POSITIVE LOGITS
     Appropriations
    0.91
    Enlarge
    0.76
     Merrill
    0.75
     Mellon
    0.73
     Commissioners
    0.72
     Press
    0.71
     accredited
    0.70
     DeVos
    0.70
     Framework
    0.69
     Examination
    0.67
    Activations Density 0.065%

    No Known Activations