© 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 3072
    9. 8-RES_FS3072-JB
    11. 244
    Prev
    Next
    INDEX
    Explanations

    details related to police investigations and requests for information

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs3072-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    3,072
    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
    726
    +0.13
    0.6%
    639
    +0.11
    0.5%
    5
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    726
    +0.13
    0.08
    639
    +0.11
    0.06
    380
    +0.10
    0.05
    Negative Logits
     stacked
    -0.81
    rawdownloadcloneembedreportprint
    -0.75
     stacking
    -0.74
     dominates
    -0.70
     slices
    -0.70
    çĶŁ
    -0.70
     dominating
    -0.69
    puff
    -0.69
     reused
    -0.69
     chunks
    -0.68
    POSITIVE LOGITS
     hotline
    1.50
     Contact
    1.06
    contact
    1.05
    Fax
    1.03
     inquiries
    1.03
     Telephone
    1.02
     enqu
    1.02
     contact
    1.01
    email
    1.00
    Email
    1.00
    Activations Density 1.543%

    No Known Activations