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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlogSlackPrivacy & 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. 46162
    Prev
    Next
    INDEX
    Explanations

    references to geographical locations in the southern region

    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
    496
    +0.16
    0.7%
    238
    +0.14
    0.7%
    314
    +0.12
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    238
    +0.16
    0.03
    596
    +0.14
    0.03
    496
    +0.12
    0.03
    Negative Logits
    roma
    -0.73
     MFT
    -0.70
    ãĤ¤ãĥĪ
    -0.69
    ICLE
    -0.68
     Unch
    -0.68
    elsen
    -0.67
    brance
    -0.67
     Finder
    -0.65
    Legendary
    -0.65
    ause
    -0.64
    POSITIVE LOGITS
    ward
    1.21
    wark
    1.17
    side
    0.99
    wards
    0.97
    west
    0.95
    bound
    0.94
    western
    0.91
    ampton
    0.88
    overs
    0.85
     east
    0.84
    Activations Density 0.017%

    No Known Activations