© 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. 23009
    Prev
    Next
    INDEX
    Explanations

    references to evidence and its sufficiency in various contexts

    oai_token-act-pair · gpt-4o-miniTriggered 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
    272
    +0.12
    0.5%
    447
    +0.10
    0.4%
    666
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    272
    +0.12
    0.03
    114
    +0.10
    0.03
    435
    +0.09
    0.02
    Negative Logits
    iery
    -0.76
     throats
    -0.71
    aeper
    -0.70
    skill
    -0.70
    ategory
    -0.68
    ttle
    -0.67
    cers
    -0.67
    cer
    -0.66
    Hop
    -0.65
    aughs
    -0.64
    POSITIVE LOGITS
     evidence
    0.99
    evidence
    0.97
     linking
    0.96
     suggesting
    0.93
     corrobor
    0.91
     indicating
    0.90
     contradict
    0.88
     conclusive
    0.87
     suggests
    0.87
     demonstrating
    0.87
    Activations Density 0.034%

    No Known Activations