© 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 12288
    9. 8-RES_FS12288-JB
    11. 1353
    Prev
    Next
    INDEX
    Explanations

    technical terms related to data security and internet technology

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    540
    +0.14
    0.7%
    206
    +0.12
    0.6%
    706
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    206
    +0.14
    0.03
    540
    +0.12
    0.03
    70
    +0.11
    0.03
    Negative Logits
    raft
    -0.74
    ship
    -0.72
    brow
    -0.68
    inki
    -0.66
    issue
    -0.64
    ï¸ı
    -0.64
    fram
    -0.63
    govtrack
    -0.63
    spr
    -0.62
     Fog
    -0.61
    POSITIVE LOGITS
     abortions
    0.90
     evaluations
    0.86
     poorly
    0.85
     miracles
    0.83
     CPR
    0.82
     duties
    0.77
    ative
    0.76
     stunts
    0.74
    uate
    0.73
     err
    0.72
    Activations Density 0.075%

    No Known Activations