© 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. 10302
    Prev
    Next
    INDEX
    Explanations

    terms related to mathematical statistics and software development, particularly focusing on specific programming languages and skills

    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
    +0.45
    2.2%
    642
    +0.13
    0.6%
    266
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    480
    +0.45
    0.02
    266
    +0.13
    0.10
    642
    +0.11
    0.04
    Negative Logits
    pload
    -0.80
    ãĥ¼ãĥ³
    -0.72
     staggered
    -0.67
    ardless
    -0.61
    essage
    -0.60
     Rebell
    -0.60
     consecut
    -0.59
     territ
    -0.59
     withdrawn
    -0.58
    secut
    -0.58
    POSITIVE LOGITS
     sciences
    0.90
     anthropology
    0.86
     academia
    0.85
    icist
    0.83
     disciplines
    0.76
    science
    0.75
     science
    0.74
     humanities
    0.74
    etics
    0.73
     philosophers
    0.73
    Activations Density 1.470%

    No Known Activations