© 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 12288
    9. 8-RES_FS12288-JB
    11. 540
    Prev
    Next
    INDEX
    Explanations

    questions starting with "What would" or "What if"

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    449
    +0.10
    0.5%
    308
    +0.09
    0.4%
    305
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    449
    +0.10
    0.05
    287
    +0.09
    0.04
    638
    +0.09
    0.04
    Negative Logits
    fort
    -0.83
    cule
    -0.73
    cius
    -0.70
    mos
    -0.67
    sha
    -0.65
    Maker
    -0.65
    xon
    -0.64
    gra
    -0.61
    Desk
    -0.61
    ements
    -0.60
    POSITIVE LOGITS
     happen
    1.15
     happens
    1.10
     happened
    1.08
     happ
    1.04
     transpired
    1.03
     awaits
    0.86
     constitutes
    0.86
     Happ
    0.84
     happening
    0.84
     distinguishes
    0.75
    Activations Density 0.142%

    No Known Activations