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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 4285
    Prev
    Next
    INDEX
    Explanations

    phrases related to physical exertion or emotional distress

    oai_token-act-pair · gpt-3.5-turboTriggered 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
    373
    +0.11
    0.5%
    634
    +0.10
    0.5%
    760
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    566
    +0.11
    0.05
    762
    +0.10
    0.06
    634
    +0.09
    0.05
    Negative Logits
     backdrop
    -0.77
     understandably
    -0.76
     reportedly
    -0.71
     prominently
    -0.65
     doubtless
    -0.65
    wik
    -0.64
     predictably
    -0.64
     ensures
    -0.64
     cite
    -0.63
     evidently
    -0.63
    POSITIVE LOGITS
    acea
    0.85
    better
    0.75
    poke
    0.72
    omething
    0.72
    enough
    0.70
    thood
    0.69
     invincible
    0.69
     kindred
    0.68
    changed
    0.66
    tin
    0.65
    Activations Density 0.423%

    No Known Activations