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 1536
    5. 8-RES_FS1536-JB
    6. 422
    Prev
    Next
    INDEX
    Explanations

    phrases related to governmental and political activities, such as discussions, appointments, and alliances

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs1536-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    1,536
    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
    200
    +0.12
    0.6%
    21
    +0.10
    0.5%
    682
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    146
    +0.12
    0.12
    288
    +0.10
    0.15
    224
    +0.09
    0.12
    Negative Logits
    oise
    -0.77
    amaz
    -0.77
    iser
    -0.73
    gra
    -0.72
    Interstitial
    -0.72
    #$#$
    -0.66
    esm
    -0.64
    GO
    -0.64
    oven
    -0.62
     Samurai
    -0.62
    POSITIVE LOGITS
     aide
    1.09
     aides
    1.08
     secretaries
    1.04
     memos
    1.03
     briefings
    0.98
     counsel
    0.98
     staffers
    0.97
     memo
    0.97
     staffer
    0.97
     adviser
    0.95
    Activations Density 2.460%

    No Known Activations