© 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. 535
    Prev
    Next
    INDEX
    Explanations

    US states and specific public policies

    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
    +0.26
    1.3%
    481
    +0.22
    1.1%
    67
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    67
    +0.26
    0.10
    65
    +0.22
    0.08
    461
    +0.12
    0.06
    Negative Logits
    akable
    -0.75
     Emin
    -0.66
    ãģĨ
    -0.65
    naissance
    -0.64
     Clair
    -0.63
    Reilly
    -0.60
    pring
    -0.59
    vous
    -0.59
    ////
    -0.58
    ãĥł
    -0.57
    POSITIVE LOGITS
     residents
    0.96
    ans
    0.93
     natives
    0.91
     policymakers
    0.90
     citizens
    0.90
    ians
    0.90
     taxpayers
    0.88
     Pradesh
    0.87
     voters
    0.84
     politicians
    0.81
    Activations Density 0.426%

    No Known Activations