© 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
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 12288
    5. 8-RES_FS12288-JB
    6. 322
    Prev
    Next
    INDEX
    Explanations

    phrases indicating future intentions or preferences

    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
    566
    +0.12
    0.5%
    502
    +0.11
    0.5%
    49
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    49
    +0.12
    0.09
    551
    +0.11
    0.08
    566
    +0.11
    0.08
    Negative Logits
     Borders
    -0.59
     Fairfax
    -0.58
     Flex
    -0.57
     Pak
    -0.57
    )=(
    -0.57
     Forensic
    -0.56
    Independent
    -0.56
     Hastings
    -0.56
     surges
    -0.55
     largeDownload
    -0.54
    POSITIVE LOGITS
     deem
    0.95
     accomplish
    0.80
     encounter
    0.78
     emulate
    0.77
     cherish
    0.72
     inhabit
    0.70
     achieve
    0.70
     termed
    0.69
     dearly
    0.69
     aspire
    0.66
    Activations Density 0.325%

    No Known Activations