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 49152
    5. 8-RES_FS49152-JB
    6. 45736
    Prev
    Next
    INDEX
    Explanations

    terms related to disputes, complaints, and issues

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    447
    +0.17
    0.8%
    288
    +0.11
    0.5%
    16
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    575
    +0.17
    0.06
    288
    +0.11
    0.05
    78
    +0.10
    0.04
    Negative Logits
     corridors
    -0.79
     fixtures
    -0.75
     regions
    -0.73
     channels
    -0.72
     brackets
    -0.71
     strains
    -0.70
     departments
    -0.70
     charts
    -0.70
    iaries
    -0.70
     practices
    -0.69
    POSITIVE LOGITS
     somewhere
    0.87
     someday
    0.81
     anywhere
    0.80
     WITHOUT
    0.78
     like
    0.78
     thats
    0.77
    cko
    0.77
     resembling
    0.76
     Called
    0.74
     anymore
    0.71
    Activations Density 0.183%

    No Known Activations