© 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. 20073
    Prev
    Next
    INDEX
    Explanations

    reflexive pronouns followed by a verb, assumingly expressing self-related actions and emotions

    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
    217
    +0.12
    0.5%
    197
    +0.11
    0.5%
    125
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    125
    +0.12
    0.04
    761
    +0.11
    0.04
    217
    +0.10
    0.03
    Negative Logits
    Zip
    -0.64
    vice
    -0.64
     Clicker
    -0.63
    letters
    -0.61
    FK
    -0.60
     Dragons
    -0.58
    BIL
    -0.56
    rats
    -0.55
     PHI
    -0.54
    ORY
    -0.54
    POSITIVE LOGITS
     by
    1.44
    by
    1.09
     BY
    0.99
     By
    0.81
    upon
    0.80
    pez
    0.79
    By
    0.78
     aback
    0.77
     via
    0.74
     elsewhere
    0.74
    Activations Density 0.152%

    No Known Activations