Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 1536
    5. 8-RES_FS1536-JB
    6. 944
    Prev
    Next
    INDEX
    Explanations

    references to financial transactions, particularly involving buying or purchasing

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs1536-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    1,536
    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
    123
    +0.10
    0.5%
    31
    +0.09
    0.4%
    525
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    665
    +0.10
    0.09
    570
    +0.09
    0.02
    123
    +0.09
    0.07
    Negative Logits
    icho
    -0.80
    pmwiki
    -0.77
    icone
    -0.68
    iquette
    -0.68
    xual
    -0.67
    Tip
    -0.67
    ogen
    -0.65
    ureau
    -0.64
    dayName
    -0.63
    erella
    -0.63
    POSITIVE LOGITS
     groceries
    1.06
     tickets
    0.97
     goods
    0.89
     pricey
    0.85
     cheaply
    0.82
     shares
    0.81
     licences
    0.81
     bitcoins
    0.81
     cheap
    0.80
     certificates
    0.79
    Activations Density 1.842%

    No Known Activations