© 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 24576
    9. 8-RES_FS24576-JB
    11. 18846
    Prev
    Next
    INDEX
    Explanations

    verbs in the past tense and descriptions of actions happening in the past

    oai_token-act-pair · gpt-3.5-turboTriggered 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
    36
    +0.22
    1.1%
    362
    +0.12
    0.6%
    730
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    36
    +0.22
    0.12
    362
    +0.12
    0.07
    250
    +0.09
    0.04
    Negative Logits
     assessments
    -0.83
    votes
    -0.82
    Events
    -0.81
     estimates
    -0.80
     indications
    -0.76
    agents
    -0.76
     answers
    -0.76
     reasons
    -0.75
     visas
    -0.74
    reports
    -0.74
    POSITIVE LOGITS
     pile
    0.96
     pedest
    0.96
     doorway
    0.95
     bunch
    0.95
     hole
    0.93
     snowy
    0.92
     mound
    0.91
     basket
    0.91
    rouse
    0.91
     crowded
    0.89
    Activations Density 0.216%

    No Known Activations