© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 768
    9. 8-RES_FS768-JB
    11. 683
    Prev
    Next
    INDEX
    Explanations

    references to stakes or importance in various contexts

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs768-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    768
    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
    650
    +0.14
    0.7%
    102
    +0.13
    0.6%
    458
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    102
    +0.14
    0.24
    55
    +0.13
    0.21
    650
    +0.11
    0.18
    Negative Logits
     spends
    -0.71
     houses
    -0.69
     retains
    -0.65
     Ago
    -0.65
    inia
    -0.65
     defends
    -0.65
     Flavoring
    -0.65
     watches
    -0.65
     Serving
    -0.65
     subscrib
    -0.63
    POSITIVE LOGITS
    ģ«
    0.96
     obscured
    0.94
     dictated
    0.93
     blurred
    0.92
     muted
    0.90
     dwindling
    0.89
     waning
    0.88
     evident
    0.88
     warranted
    0.86
     inex
    0.85
    Activations Density 5.376%

    No Known Activations