© 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. 18807
    Prev
    Next
    INDEX
    Explanations

    hyphenated words with negative connotations

    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
    640
    +0.12
    0.6%
    199
    +0.12
    0.5%
    314
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    199
    +0.12
    0.02
    3
    +0.12
    0.02
    616
    +0.11
    0.02
    Negative Logits
    ulhu
    -0.98
     Juliet
    -0.81
     Dickinson
    -0.75
     Ranger
    -0.72
     Bravo
    -0.71
     Bits
    -0.70
     Doe
    -0.70
    itaire
    -0.68
     Compass
    -0.68
     Guru
    -0.67
    POSITIVE LOGITS
    emphasis
    1.25
    regulation
    1.19
    sized
    1.15
    priced
    1.13
    performing
    1.10
    regulated
    1.09
    arching
    1.09
    exc
    1.09
    reg
    1.05
    reporting
    1.05
    Activations Density 0.016%

    No Known Activations