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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & 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. 45532
    Prev
    Next
    INDEX
    Explanations

    references to people's collective experiences and perspectives in discussions

    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
    393
    +0.15
    0.7%
    481
    +0.12
    0.6%
    496
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    139
    +0.15
    0.04
    354
    +0.12
    0.04
    762
    +0.11
    0.04
    Negative Logits
     Complex
    -0.63
     Fine
    -0.63
     Bound
    -0.61
     Press
    -0.61
     Ivory
    -0.60
     Xan
    -0.60
     Door
    -0.60
     Ball
    -0.60
    Kal
    -0.60
     Dangerous
    -0.59
    POSITIVE LOGITS
    're
    1.04
    've
    0.90
     concede
    0.89
     aspire
    0.88
     intend
    0.86
     stray
    0.84
     retire
    0.83
     succumb
    0.82
    afort
    0.82
    abouts
    0.81
    Activations Density 0.110%

    No Known Activations