© 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
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 12288
    5. 8-RES_FS12288-JB
    6. 10120
    Prev
    Next
    INDEX
    Explanations

    technical terms related to computer processors, anatomy, academic research, and cryptocurrency

    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
    480
    +0.39
    1.9%
    716
    +0.10
    0.5%
    628
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.39
    0.14
    616
    +0.10
    0.08
    652
    +0.08
    0.07
    Negative Logits
    ovember
    -0.69
    terday
    -0.67
     âĵĺ
    -0.67
    utics
    -0.64
    olia
    -0.63
     Rica
    -0.63
     Disorder
    -0.59
     Leone
    -0.58
    Lago
    -0.58
    ornings
    -0.57
    POSITIVE LOGITS
    iest
    1.21
    liest
    1.10
     portion
    1.05
    osphere
    1.03
     aspect
    0.89
     element
    0.88
     equivalent
    0.88
    hest
    0.85
     side
    0.82
     axis
    0.81
    Activations Density 1.543%

    No Known Activations