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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 4388
    Prev
    Next
    INDEX
    Explanations

    references to challenges or problems that need to be addressed or overcome

    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
    46
    +0.10
    0.5%
    27
    +0.10
    0.5%
    34
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    51
    +0.10
    0.03
    27
    +0.10
    0.03
    247
    +0.10
    0.03
    Negative Logits
    otide
    -0.72
    RAL
    -0.72
    azeera
    -0.71
    VO
    -0.65
    smoking
    -0.65
    ovies
    -0.64
    olphins
    -0.63
    endar
    -0.63
     snippets
    -0.63
    ools
    -0.63
    POSITIVE LOGITS
     confronting
    1.17
     facing
    1.15
     posed
    1.13
     faced
    1.11
     encountered
    0.98
     arise
    0.94
     plag
    0.93
    hooting
    0.93
     arising
    0.91
     daunting
    0.91
    Activations Density 0.062%

    No Known Activations