© 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. 642
    Prev
    Next
    INDEX
    Explanations

    pronouns and modal verbs indicating capability or possibility

    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
    208
    +0.11
    0.5%
    598
    +0.10
    0.5%
    683
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    598
    +0.11
    0.05
    566
    +0.10
    0.06
    762
    +0.10
    0.06
    Negative Logits
    undy
    -0.76
    代
    -0.65
    ixt
    -0.64
    ãģ®å
    -0.63
    è¦ļéĨĴ
    -0.62
    program
    -0.61
     Forty
    -0.60
     Manufact
    -0.60
    ornia
    -0.59
    Opening
    -0.59
    POSITIVE LOGITS
     nonetheless
    1.46
     nevertheless
    1.41
     still
    1.10
     didn
    1.09
     ain
    1.07
     doesn
    1.05
     alas
    1.04
     wouldn
    1.01
     cannot
    1.01
     couldn
    1.01
    Activations Density 0.333%

    No Known Activations