© 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 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. 12948
    Prev
    Next
    INDEX
    Explanations

    instructions or guidelines on how to perform different actions or tasks in a step-by-step manner

    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
    551
    +0.13
    0.6%
    297
    +0.12
    0.5%
    715
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    551
    +0.13
    0.08
    49
    +0.12
    0.06
    60
    +0.10
    0.05
    Negative Logits
    pires
    -0.79
     Moz
    -0.69
     Niet
    -0.65
     Hits
    -0.64
     Mour
    -0.63
     Closing
    -0.63
    bats
    -0.62
    rant
    -0.62
    cule
    -0.61
    polit
    -0.61
    POSITIVE LOGITS
    't
    1.46
     optionally
    1.25
     choose
    1.19
     customize
    1.14
    NOT
    1.12
     utilize
    1.11
     use
    1.05
     skip
    1.03
     specify
    1.02
     manipulate
    1.02
    Activations Density 0.126%

    No Known Activations