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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 24576
    9. 8-RES_FS24576-JB
    11. 22059
    Prev
    Next
    INDEX
    Explanations

    occurrences of the phrase "I have to" or similar expressions conveying obligation or necessity

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    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
    551
    +0.13
    0.6%
    502
    +0.11
    0.5%
    49
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    551
    +0.13
    0.06
    49
    +0.11
    0.06
    105
    +0.11
    0.04
    Negative Logits
     Lambert
    -0.69
     Emerging
    -0.69
    Recommended
    -0.67
     Globe
    -0.66
    furt
    -0.63
     Williamson
    -0.62
     Rust
    -0.61
    vironment
    -0.61
    closed
    -0.58
     Palest
    -0.58
    POSITIVE LOGITS
     rely
    1.18
     wait
    1.10
     obey
    1.09
     rethink
    1.07
     abide
    1.07
     prove
    1.04
     remind
    1.03
     prioritize
    1.02
     admit
    1.01
     comply
    1.01
    Activations Density 0.097%

    No Known Activations