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 49152
    5. 8-RES_FS49152-JB
    6. 43308
    Prev
    Next
    INDEX
    Explanations

    references to the Final Fantasy video game series and its various titles

    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
    288
    +0.13
    0.6%
    48
    +0.10
    0.5%
    556
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    556
    +0.13
    0.01
    359
    +0.10
    0.01
    766
    +0.09
    0.01
    Negative Logits
    lain
    -0.86
    uckland
    -0.85
    conserv
    -0.82
    ask
    -0.80
    imentary
    -0.79
    aspberry
    -0.79
    asking
    -0.78
    shr
    -0.76
    acity
    -0.75
    inventoryQuantity
    -0.71
    POSITIVE LOGITS
     XIV
    1.28
     XV
    1.24
     VII
    1.21
     XIII
    1.19
     VIII
    1.10
     IX
    1.07
     XII
    1.05
     XI
    1.02
     Tactics
    1.01
     Versus
    1.00
    Activations Density 0.012%

    No Known Activations