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 49152
    5. 8-RES_FS49152-JB
    6. 25623
    Prev
    Next
    INDEX
    Explanations

    terms related to economic crises and their impacts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    436
    +0.12
    0.5%
    756
    +0.11
    0.5%
    542
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    300
    +0.12
    0.04
    150
    +0.11
    0.03
    546
    +0.11
    0.03
    Negative Logits
     Assist
    -0.82
     salsa
    -0.80
    oyal
    -0.80
     interchangeable
    -0.74
     ears
    -0.73
    ogly
    -0.73
    ibo
    -0.71
    pees
    -0.71
    ioch
    -0.71
     clipboard
    -0.71
    POSITIVE LOGITS
     exacerbated
    1.14
     engulf
    1.07
     perpetrated
    1.05
     devast
    1.01
     rav
    1.01
     ravaged
    1.00
     uphe
    1.00
     caused
    0.99
     waged
    0.99
     raged
    0.99
    Activations Density 0.278%

    No Known Activations