Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlogSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 768
    5. 8-RES_FS768-JB
    6. 328
    Prev
    Next
    INDEX
    Explanations

    corporate information and contact details

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs768-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    768
    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
    616
    +0.09
    0.4%
    289
    +0.09
    0.4%
    129
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    326
    +0.09
    0.16
    224
    +0.09
    0.12
    511
    +0.09
    0.11
    Negative Logits
    istically
    -0.65
    aneously
    -0.61
    ries
    -0.58
    ishers
    -0.56
     judges
    -0.55
    ASED
    -0.55
    attled
    -0.54
    QB
    -0.53
     runoff
    -0.53
    istic
    -0.53
    POSITIVE LOGITS
    peed
    1.57
    hip
    1.52
    hift
    1.50
    ystem
    1.49
    chool
    1.48
    creen
    1.45
    mith
    1.45
    aurus
    1.44
    erver
    1.41
    hips
    1.38
    Activations Density 4.202%

    No Known Activations