© 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. 1
    Prev
    Next
    INDEX
    Explanations

    political and legal terms and concepts

    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
    481
    +0.14
    0.7%
    745
    +0.10
    0.5%
    98
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    96
    +0.14
    0.07
    134
    +0.10
    0.07
    516
    +0.09
    0.06
    Negative Logits
    ortality
    -0.74
     Surviv
    -0.73
     Purg
    -0.72
    knit
    -0.70
     Nanto
    -0.70
     Puzzles
    -0.67
     Lazarus
    -0.66
    ordial
    -0.65
     Sunder
    -0.65
     Fold
    -0.64
    POSITIVE LOGITS
     unfairly
    1.42
     misrepresent
    1.19
     improperly
    1.18
     unfair
    1.17
     unjust
    1.12
     mishand
    1.11
     violating
    1.11
     violated
    1.10
     tresp
    1.09
     inaccur
    1.08
    Activations Density 2.622%

    No Known Activations