© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 24576
    9. 8-RES_FS24576-JB
    11. 12026
    Prev
    Next
    INDEX
    Explanations

    proper nouns and terms related to legal procedures and various industries

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    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
    169
    +0.11
    0.5%
    76
    +0.10
    0.5%
    631
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    401
    +0.11
    0.06
    502
    +0.10
    0.04
    606
    +0.10
    0.05
    Negative Logits
    onet
    -0.80
    geries
    -0.74
    ocious
    -0.73
    raints
    -0.72
    acteria
    -0.68
    DragonMagazine
    -0.67
    maxwell
    -0.67
    acterial
    -0.65
    ailability
    -0.64
    ategory
    -0.64
    POSITIVE LOGITS
     deems
    1.32
     deem
    1.08
     sorely
    0.96
     considers
    0.95
     lacked
    0.90
     describes
    0.88
     refers
    0.87
     chooses
    0.86
     forgot
    0.85
     hates
    0.85
    Activations Density 0.239%

    No Known Activations