© 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. 843
    Prev
    Next
    INDEX
    Explanations

    information related to instructions or procedures

    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
    595
    +0.10
    0.5%
    399
    +0.10
    0.4%
    562
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    484
    +0.10
    0.05
    595
    +0.10
    0.04
    399
    +0.10
    0.04
    Negative Logits
    ocracy
    -0.82
    hed
    -0.71
    ocratic
    -0.68
    eenth
    -0.68
    opia
    -0.68
    ENE
    -0.66
    eth
    -0.65
    elong
    -0.61
    OST
    -0.61
    ocrats
    -0.60
    POSITIVE LOGITS
     alternatively
    1.59
    Else
    1.28
     else
    1.23
    chard
    1.22
    acles
    1.12
    acle
    1.08
    chid
    1.01
    nam
    1.00
    acular
    0.98
    phans
    0.97
    Activations Density 0.156%

    No Known Activations