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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsBlogSlackPrivacy & 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. 1433
    Prev
    Next
    INDEX
    Explanations

    dates and time references

    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
    525
    +0.10
    0.5%
    702
    +0.10
    0.4%
    612
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    702
    +0.10
    0.06
    337
    +0.10
    0.05
    654
    +0.09
    0.05
    Negative Logits
     inherit
    -0.68
     Genie
    -0.60
    avorite
    -0.59
     dearly
    -0.59
     accurately
    -0.59
     distortions
    -0.57
    Reviewer
    -0.56
     truly
    -0.56
     gorilla
    -0.56
    /"
    -0.54
    POSITIVE LOGITS
    imester
    0.98
    morning
    0.91
     afternoon
    0.83
    cture
    0.82
    ciating
    0.80
     evening
    0.79
    season
    0.74
     gestation
    0.74
    eteenth
    0.73
    cheon
    0.70
    Activations Density 0.773%

    No Known Activations