© 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 49152
    9. 8-RES_FS49152-JB
    11. 26570
    Prev
    Next
    INDEX
    Explanations

    mentions of Americans and their opinions or actions

    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
    +0.11
    0.5%
    282
    +0.10
    0.5%
    481
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    282
    +0.11
    0.02
    254
    +0.10
    0.02
    193
    +0.10
    0.02
    Negative Logits
    operation
    -0.64
    nec
    -0.64
     Dj
    -0.62
    ces
    -0.62
    NB
    -0.62
    cer
    -0.62
     Samson
    -0.61
    Initialized
    -0.61
    Prev
    -0.61
    clusive
    -0.60
    POSITIVE LOGITS
    ourcing
    0.88
    icans
    0.83
    hip
    0.80
    chool
    0.80
    ourced
    0.79
    ervatives
    0.76
    ophobia
    0.74
     Skies
    0.74
    ervative
    0.73
    aurus
    0.73
    Activations Density 0.025%

    No Known Activations