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. 26195
    Prev
    Next
    INDEX
    Explanations

    verbs representing actions or activities

    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
    38
    +0.11
    0.5%
    157
    +0.10
    0.5%
    659
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    70
    +0.11
    0.06
    665
    +0.10
    0.05
    427
    +0.10
    0.03
    Negative Logits
     destro
    -0.67
     streng
    -0.63
    ":"/
    -0.60
    ophy
    -0.59
     Ange
    -0.58
    geant
    -0.57
     Persons
    -0.56
    issance
    -0.56
     msec
    -0.56
    ãĤ±
    -0.56
    POSITIVE LOGITS
    ingly
    0.98
    eth
    0.89
    fully
    0.86
     themselves
    0.85
    ings
    0.85
    adelphia
    0.84
    ned
    0.79
    backs
    0.75
    ively
    0.71
    selves
    0.70
    Activations Density 0.302%

    No Known Activations