© 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. 9685
    Prev
    Next
    INDEX
    Explanations

    superlatives, specifically finding the word "most" in various contexts

    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
    190
    +0.11
    0.5%
    726
    +0.11
    0.5%
    102
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    190
    +0.11
    0.07
    194
    +0.11
    0.06
    726
    +0.11
    0.06
    Negative Logits
    æ©
    -0.78
    pload
    -0.70
     Creed
    -0.70
    heid
    -0.69
    eto
    -0.68
    instead
    -0.67
    ategories
    -0.67
     Films
    -0.66
    fell
    -0.66
    halla
    -0.64
    POSITIVE LOGITS
     important
    1.17
     powerful
    1.10
     expensive
    1.08
     exciting
    1.05
     efficient
    1.04
     interesting
    1.03
     valuable
    1.02
     recognizable
    1.01
     lucrative
    1.00
     profitable
    1.00
    Activations Density 0.098%

    No Known Activations