© 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. 12142
    Prev
    Next
    INDEX
    Explanations

    punctuation marks, specifically parentheses and dashes

    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
    481
    +0.20
    0.9%
    278
    +0.10
    0.5%
    291
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    698
    +0.20
    0.04
    517
    +0.10
    0.03
    537
    +0.10
    0.03
    Negative Logits
     Betsy
    -0.58
     PSU
    -0.56
    ibl
    -0.56
     Cyn
    -0.54
     SJ
    -0.54
     cynicism
    -0.53
    clus
    -0.53
     blaster
    -0.53
    sonian
    -0.53
     Omaha
    -0.52
    POSITIVE LOGITS
     belonging
    0.90
    into
    0.89
    containing
    0.87
    without
    0.85
    necessary
    0.76
     onto
    0.73
    along
    0.73
    imb
    0.73
     interchange
    0.73
     pertaining
    0.72
    Activations Density 0.122%

    No Known Activations