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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 3072
    5. 8-RES_FS3072-JB
    6. 2224
    Prev
    Next
    INDEX
    Explanations

    phrases related to education and training

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs3072-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    3,072
    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
    447
    +0.17
    0.8%
    481
    +0.12
    0.6%
    330
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    330
    +0.17
    0.12
    224
    +0.12
    0.12
    32
    +0.11
    0.12
    Negative Logits
    Accessory
    -0.67
    Hig
    -0.65
    sole
    -0.64
    clusive
    -0.63
    Aw
    -0.61
    Ore
    -0.61
    mone
    -0.60
    Translation
    -0.60
    Planet
    -0.60
    Redd
    -0.60
    POSITIVE LOGITS
     beware
    1.17
     are
    1.16
     aren
    1.15
     reacted
    1.08
     cannot
    1.06
     weren
    1.03
    hip
    1.03
     rejoice
    1.03
     were
    1.01
     alike
    1.01
    Activations Density 1.306%

    No Known Activations