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 1536
    5. 8-RES_FS1536-JB
    6. 1186
    Prev
    Next
    INDEX
    Explanations

    references to politics, sports, and personal achievements

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs1536-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    1,536
    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
    57
    +0.12
    0.6%
    624
    +0.12
    0.5%
    631
    +0.12
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    57
    +0.12
    0.08
    659
    +0.12
    0.09
    631
    +0.12
    0.09
    Negative Logits
     Anthem
    -0.80
     Kitchen
    -0.66
    ãĤ¤ãĥĪ
    -0.63
    iets
    -0.63
    ivan
    -0.62
    rans
    -0.60
     Painter
    -0.60
    rooms
    -0.60
     Bleach
    -0.60
     Vulkan
    -0.60
    POSITIVE LOGITS
    gypt
    1.23
    lements
    1.16
    tymology
    1.13
     plur
    1.11
    ternally
    1.10
    ighty
    1.09
    uphem
    1.08
    agles
    1.07
    instein
    1.04
    cosystem
    1.04
    Activations Density 1.401%

    No Known Activations