© 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. 37253
    Prev
    Next
    INDEX
    Explanations

    expressions of personal opinion and subjective analysis

    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
    87
    +0.33
    1.7%
    464
    +0.14
    0.7%
    481
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    87
    +0.33
    -0.01
    393
    +0.14
    0.06
    268
    +0.09
    0.02
    Negative Logits
    ufact
    -0.77
    eworks
    -0.72
     finished
    -0.69
    finished
    -0.66
    ascus
    -0.65
     TODAY
    -0.65
     comply
    -0.62
     finishes
    -0.62
    Congratulations
    -0.61
    otte
    -0.61
    POSITIVE LOGITS
     pervasive
    0.84
     tendency
    0.83
     alienation
    0.80
     loneliness
    0.79
     paradox
    0.78
     paranoia
    0.77
     perme
    0.77
     narciss
    0.74
     perception
    0.74
     coined
    0.74
    Activations Density 0.412%

    No Known Activations