© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 49152
    9. 8-RES_FS49152-JB
    11. 41735
    Prev
    Next
    INDEX
    Explanations

    punctuation marks, particularly commas

    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
    765
    +0.10
    0.5%
    393
    +0.10
    0.5%
    347
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    765
    +0.10
    0.07
    546
    +0.10
    0.06
    43
    +0.10
    0.05
    Negative Logits
    ¬¼
    -0.63
     advers
    -0.62
    iece
    -0.61
    ģĸ
    -0.61
     adversity
    -0.61
     wart
    -0.60
    ĨĴ
    -0.59
    Į
    -0.58
    KEN
    -0.57
     pandemonium
    -0.57
    POSITIVE LOGITS
     albeit
    1.07
     including
    1.00
     irrespective
    0.96
     regardless
    0.94
    including
    0.93
     Including
    0.93
     preferably
    0.92
    mostly
    0.92
     although
    0.88
     mainly
    0.87
    Activations Density 0.218%

    No Known Activations