© 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. 41814
    Prev
    Next
    INDEX
    Explanations

    patterns of non-Latin characters and possibly language identifiers

    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
    730
    +0.10
    0.5%
    124
    +0.10
    0.4%
    116
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    289
    +0.10
    0.00
    124
    +0.10
    0.01
    730
    +0.09
    0.01
    Negative Logits
     Fargo
    -0.71
     WARN
    -0.66
    wagen
    -0.65
     downs
    -0.64
     Glory
    -0.64
     computing
    -0.63
     recall
    -0.63
     bailout
    -0.62
     conception
    -0.61
     greed
    -0.61
    POSITIVE LOGITS
    Ĩ
    1.54
    Ĭ
    1.53
    ĩ
    1.47
    ģ
    1.30
    Ĥ
    1.25
    ĥ
    1.21
    ¦
    1.11
    ÙĦ
    1.08
    º
    1.06
    ±
    1.06
    Activations Density 0.003%

    No Known Activations