© 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. 14331
    Prev
    Next
    INDEX
    Explanations

    phrases related to instructions or rules

    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
    480
    +0.23
    1.1%
    492
    +0.12
    0.6%
    167
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    492
    +0.23
    0.04
    49
    +0.12
    0.04
    736
    +0.11
    0.03
    Negative Logits
     Gamb
    -0.68
    forth
    -0.67
    ipal
    -0.66
     Cornwall
    -0.61
     Filip
    -0.61
     MacArthur
    -0.60
     Canaver
    -0.58
    ulla
    -0.58
    é¾
    -0.58
     Chap
    -0.58
    POSITIVE LOGITS
    're
    1.35
    've
    1.19
    'll
    1.14
    tub
    1.06
     guys
    1.06
    tu
    1.05
    'd
    0.98
    RS
    0.96
     wanna
    0.93
    kai
    0.89
    Activations Density 0.073%

    No Known Activations