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 3072
    5. 8-RES_FS3072-JB
    6. 528
    Prev
    Next
    INDEX
    Explanations

    economic and political context or commentary

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs3072-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    3,072
    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
    87
    +0.29
    1.6%
    756
    +0.22
    1.2%
    86
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    756
    +0.29
    -0.05
    87
    +0.22
    -0.08
    86
    +0.10
    0.10
    Negative Logits
     Poké
    -0.63
    )",
    -0.58
    ookie
    -0.58
     Solo
    -0.58
    Legend
    -0.57
     haha
    -0.57
    oji
    -0.56
     Completed
    -0.56
     filming
    -0.56
    VERTISEMENT
    -0.55
    POSITIVE LOGITS
     neocons
    0.97
     repud
    0.92
     democracies
    0.91
     trillions
    0.90
     progressives
    0.89
     scapego
    0.89
     neoliberal
    0.86
     ideologically
    0.85
     impover
    0.83
     liberals
    0.82
    Activations Density 6.941%

    No Known Activations