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. 43306
    Prev
    Next
    INDEX
    Explanations

    expressions related to significant negative experiences and emotional 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
    480
    +0.29
    1.4%
    481
    +0.15
    0.7%
    447
    +0.14
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    415
    +0.29
    0.05
    762
    +0.15
    0.07
    542
    +0.14
    0.05
    Negative Logits
    pires
    -0.77
    usalem
    -0.62
    pmwiki
    -0.60
    thood
    -0.60
    fortune
    -0.56
    ĪĴ
    -0.55
    CAN
    -0.55
    ãĥķãĤ¡
    -0.54
    conservancy
    -0.53
    ãĥĪ
    -0.53
    POSITIVE LOGITS
     though
    0.89
     however
    0.84
     because
    0.81
     until
    0.79
    ;
    0.78
     tho
    0.78
     but
    0.77
     whereas
    0.74
     beforehand
    0.73
     except
    0.73
    Activations Density 0.591%

    No Known Activations