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. 1847
    Prev
    Next
    INDEX
    Explanations

    times expressed in the format hour:minute am/pm

    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
    373
    +0.17
    0.8%
    111
    +0.11
    0.5%
    601
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    111
    +0.17
    0.06
    373
    +0.11
    -0.01
    601
    +0.10
    0.05
    Negative Logits
     acknowled
    -0.64
     evolution
    -0.60
     Mechdragon
    -0.60
     unintended
    -0.57
     spo
    -0.57
     rules
    -0.54
    æ³
    -0.54
     AGA
    -0.53
    ahime
    -0.53
     Leilan
    -0.53
    POSITIVE LOGITS
    clock
    0.89
    pm
    0.86
    noon
    0.83
    m
    0.82
    Downloadha
    0.79
     noon
    0.78
    conom
    0.78
    local
    0.78
     curfew
    0.77
     CST
    0.77
    Activations Density 0.322%

    No Known Activations