© 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. 874
    Prev
    Next
    INDEX
    Explanations

    measurements and units of measurement

    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
    314
    +0.13
    0.6%
    187
    +0.11
    0.5%
    260
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    545
    +0.13
    0.03
    260
    +0.11
    0.04
    392
    +0.10
    0.04
    Negative Logits
     nonexistent
    -0.63
     Siber
    -0.61
     Allah
    -0.61
     Allaah
    -0.60
     Dull
    -0.57
     Eid
    -0.56
     Niet
    -0.56
     counseling
    -0.55
     counselor
    -0.55
     501
    -0.55
    POSITIVE LOGITS
    illion
    0.96
    hz
    0.93
    +,
    0.88
    +.
    0.86
    +
    0.86
    +)
    0.85
     (£
    0.84
    £
    0.82
    bps
    0.80
    /-
    0.78
    Activations Density 0.093%

    No Known Activations