© 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. 26550
    Prev
    Next
    INDEX
    Explanations

    numerical data related to contact information and addresses

    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
    477
    +0.11
    0.5%
    734
    +0.10
    0.5%
    411
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    411
    +0.11
    0.02
    477
    +0.10
    0.02
    326
    +0.10
    0.01
    Negative Logits
    bread
    -0.68
     irresistible
    -0.63
     gravy
    -0.63
     paved
    -0.62
     perfected
    -0.61
     consistency
    -0.61
    ¿½
    -0.60
     chunk
    -0.60
     weakened
    -0.59
    Ĥİ
    -0.59
    POSITIVE LOGITS
     978
    0.94
    Fax
    0.81
     CLR
    0.74
    76561
    0.72
     Advis
    0.71
     503
    0.69
     000
    0.69
     fax
    0.67
     Email
    0.67
     Downloads
    0.67
    Activations Density 0.019%

    No Known Activations