© 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. 4086
    Prev
    Next
    INDEX
    Explanations

    descriptions of criminal activities and legal investigations

    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
    481
    +0.23
    1.1%
    679
    +0.17
    0.8%
    430
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    6
    +0.23
    0.14
    679
    +0.17
    0.13
    440
    +0.12
    0.11
    Negative Logits
     independ
    -0.74
     uncond
    -0.68
     listener
    -0.68
     tyr
    -0.68
     whis
    -0.67
     accommodation
    -0.67
     intermediate
    -0.66
     yell
    -0.66
     cabbage
    -0.65
     starvation
    -0.65
    POSITIVE LOGITS
     Indeed
    1.28
     Moreover
    1.23
     Though
    1.20
     Nonetheless
    1.19
     Nevertheless
    1.19
     Instead
    1.17
     Rather
    1.16
     Whereas
    1.13
     Unlike
    1.11
     Particularly
    1.10
    Activations Density 0.541%

    No Known Activations