Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit 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. 480
    Prev
    Next
    INDEX
    Explanations

    phrases related to locations and directions

    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
    663
    +0.10
    0.5%
    314
    +0.09
    0.4%
    95
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    298
    +0.10
    0.08
    608
    +0.09
    0.11
    362
    +0.09
    0.10
    Negative Logits
     @@
    -0.67
    %%%%
    -0.63
     Theft
    -0.55
    vich
    -0.55
     Spoiler
    -0.54
     smokes
    -0.54
     Bite
    -0.54
    together
    -0.53
     CODE
    -0.53
    wra
    -0.53
    POSITIVE LOGITS
     extent
    1.52
    venge
    1.36
     detriment
    1.36
     tune
    1.11
     fullest
    1.06
    rouse
    1.04
     liking
    0.94
     depths
    0.92
     destinations
    0.89
     conclusion
    0.88
    Activations Density 2.159%

    No Known Activations