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 49152
    5. 8-RES_FS49152-JB
    6. 43163
    Prev
    Next
    INDEX
    Explanations

    expressions of praise or acknowledgment

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    526
    +0.13
    0.6%
    314
    +0.12
    0.5%
    641
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    424
    +0.13
    0.01
    641
    +0.12
    0.01
    259
    +0.11
    0.01
    Negative Logits
     Skydragon
    -0.67
     turf
    -0.66
     Sinai
    -0.63
    Sham
    -0.63
     conflic
    -0.60
    shr
    -0.60
    udder
    -0.59
     reorgan
    -0.59
     captivity
    -0.58
     backups
    -0.58
    POSITIVE LOGITS
    ably
    1.15
    eting
    1.01
    naire
    0.94
    orial
    0.91
    ciation
    0.88
    ing
    0.88
    ager
    0.84
    aires
    0.83
    edly
    0.82
    ements
    0.81
    Activations Density 0.012%

    No Known Activations