© 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. 388
    Prev
    Next
    INDEX
    Explanations

    words related to politics, specifically government officials such as governors and presidents, as well as terms related to government activities and corruption

    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
    447
    +0.13
    0.6%
    679
    +0.12
    0.5%
    519
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    224
    +0.13
    0.08
    657
    +0.12
    0.07
    401
    +0.11
    0.07
    Negative Logits
    INGTON
    -0.72
     Territories
    -0.72
    asca
    -0.70
     Tropical
    -0.68
     Virus
    -0.65
    okemon
    -0.62
    izoph
    -0.61
    Operation
    -0.61
     Seah
    -0.59
    ¥µ
    -0.59
    POSITIVE LOGITS
    hips
    1.63
    hip
    1.62
    ervatives
    1.06
    mith
    1.06
    paces
    1.05
     who
    0.99
    who
    0.94
    esses
    0.93
    '
    0.92
    doms
    0.91
    Activations Density 0.407%

    No Known Activations