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. 1371
    Prev
    Next
    INDEX
    Explanations

    adjectives indicating intensity or severity

    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
    68
    +0.12
    0.6%
    362
    +0.12
    0.6%
    357
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    669
    +0.12
    0.11
    408
    +0.12
    0.14
    362
    +0.11
    0.13
    Negative Logits
    ften
    -0.87
    ologue
    -0.86
    undrum
    -0.85
    ittee
    -0.83
    hander
    -0.82
    washer
    -0.80
    hyde
    -0.78
    orah
    -0.77
    agonist
    -0.76
    ressor
    -0.75
    POSITIVE LOGITS
     amounts
    1.45
     quantities
    1.36
     doses
    1.23
     versions
    1.20
     levels
    1.19
     situations
    1.16
     relationships
    1.12
     intervals
    1.11
     considerations
    1.11
     proportions
    1.10
    Activations Density 1.849%

    No Known Activations