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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 24576
    5. 8-RES_FS24576-JB
    6. 8891
    Prev
    Next
    INDEX
    Explanations

    references to sports teams, particularly the Los Angeles Dodgers

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    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
    167
    +0.15
    0.7%
    432
    +0.10
    0.5%
    52
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    628
    +0.15
    0.01
    232
    +0.10
    0.01
    326
    +0.10
    0.01
    Negative Logits
    inventoryQuantity
    -0.79
     Kand
    -0.66
    ashtra
    -0.65
    olitan
    -0.65
     ILCS
    -0.65
    ropolitan
    -0.65
    _>
    -0.65
    urities
    -0.64
    HAEL
    -0.64
    terday
    -0.62
    POSITIVE LOGITS
    sworth
    1.14
    sonian
    1.11
    yssey
    1.10
    gers
    1.03
    gement
    1.02
    der
    1.01
    gins
    1.01
    ging
    1.00
    gin
    0.97
    son
    0.96
    Activations Density 0.018%

    No Known Activations