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. 1965
    Prev
    Next
    INDEX
    Explanations

    phrases related to sports drafts, particularly mentioning round numbers

    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
    157
    +0.14
    0.6%
    302
    +0.12
    0.5%
    314
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.14
    0.07
    500
    +0.12
    0.05
    302
    +0.10
    0.05
    Negative Logits
    iments
    -0.78
    enez
    -0.67
     Hots
    -0.67
    ufact
    -0.66
    hered
    -0.65
    earch
    -0.64
    AMI
    -0.64
    uras
    -0.64
     Fas
    -0.64
    yer
    -0.63
    POSITIVE LOGITS
    trip
    1.04
     rob
    1.02
    about
    1.02
    table
    1.01
    abouts
    0.93
    outs
    0.83
    tri
    0.80
    ups
    0.77
    stakes
    0.76
    ãĥĥãĥĪ
    0.72
    Activations Density 0.822%

    No Known Activations