© 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 24576
    9. 8-RES_FS24576-JB
    11. 9927
    Prev
    Next
    INDEX
    Explanations

    frequencies or numeric values within a document

    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
    588
    +0.13
    0.6%
    171
    +0.10
    0.5%
    675
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    588
    +0.13
    0.06
    384
    +0.10
    0.04
    171
    +0.10
    0.05
    Negative Logits
    ccording
    -0.74
    76561
    -0.73
    iring
    -0.70
    kept
    -0.69
     Azerb
    -0.67
    everal
    -0.65
    LOCK
    -0.65
    âĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢ
    -0.64
    ĵĺ
    -0.64
    Attempt
    -0.63
    POSITIVE LOGITS
     CST
    0.77
     PST
    0.69
     PG
    0.64
    witz
    0.61
     vice
    0.60
     HF
    0.59
     mids
    0.57
     CPS
    0.57
     tech
    0.56
     ours
    0.56
    Activations Density 0.692%

    No Known Activations