© 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. 75
    Prev
    Next
    INDEX
    Explanations

    web links, email addresses, and technical data

    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
    412
    +0.13
    0.6%
    469
    +0.12
    0.5%
    686
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    373
    +0.13
    -0.04
    412
    +0.12
    0.06
    176
    +0.11
    0.04
    Negative Logits
    izoph
    -0.72
    ynes
    -0.59
     passionate
    -0.59
     brim
    -0.58
     once
    -0.58
     ado
    -0.57
     suffice
    -0.57
     garn
    -0.56
    ignment
    -0.55
    ggles
    -0.55
    POSITIVE LOGITS
    isSpecialOrderable
    0.91
    к
    0.77
    ickey
    0.69
    âĸĪâĸĪâĸĪâĸĪ
    0.68
    1969
    0.68
    schild
    0.67
    2001
    0.67
    1000
    0.66
    370
    0.65
    2018
    0.65
    Activations Density 0.866%

    No Known Activations