© 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. 22028
    Prev
    Next
    INDEX
    Explanations

    phrases that indicate difficulty or effort in accomplishing tasks

    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
    516
    +0.12
    0.6%
    218
    +0.12
    0.5%
    675
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    351
    +0.12
    0.02
    221
    +0.12
    0.02
    516
    +0.11
    0.02
    Negative Logits
    estern
    -0.68
    aire
    -0.68
    undrum
    -0.68
     ..............
    -0.64
     ASA
    -0.64
    ãģ¯
    -0.63
     Christy
    -0.62
    geist
    -0.62
     Vald
    -0.59
    oret
    -0.59
    POSITIVE LOGITS
     necessarily
    0.90
     oneself
    0.84
     slightest
    0.78
     explicitly
    0.77
     whatsoever
    0.75
     anymore
    0.70
     beforehand
    0.69
     expressly
    0.66
     consciously
    0.66
     anything
    0.66
    Activations Density 0.114%

    No Known Activations