Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 1536
    5. 8-RES_FS1536-JB
    6. 503
    Prev
    Next
    INDEX
    Explanations

    various opinions and discussions related to religious beliefs, political issues, and social topics

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs1536-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    1,536
    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
    562
    +0.12
    0.5%
    186
    +0.10
    0.5%
    762
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    8
    +0.12
    0.10
    718
    +0.10
    0.11
    119
    +0.10
    0.08
    Negative Logits
    icipated
    -0.73
     Roll
    -0.56
    riers
    -0.55
     Crush
    -0.54
     Skydragon
    -0.54
     buildup
    -0.52
    drawn
    -0.51
     valued
    -0.50
    Winged
    -0.50
    ership
    -0.49
    POSITIVE LOGITS
     yes
    1.08
    YES
    1.06
     answer
    1.00
    Yes
    1.00
     anyway
    0.99
     Nope
    0.98
    yes
    0.96
    laughs
    0.95
     YES
    0.91
    Answer
    0.91
    Activations Density 3.814%

    No Known Activations