© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 768
    9. 8-RES_FS768-JB
    11. 685
    Prev
    Next
    INDEX
    Explanations

    verb phrases preceded by 'Here is' or 'Here are'

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs768-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    768
    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
    377
    +0.11
    0.5%
    276
    +0.10
    0.5%
    651
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    377
    +0.11
    0.11
    651
    +0.10
    0.09
    276
    +0.10
    0.09
    Negative Logits
    ahime
    -0.56
    llor
    -0.56
     warranties
    -0.56
    ãĤ£
    -0.56
    inventoryQuantity
    -0.55
     Thrones
    -0.54
     Breath
    -0.54
    Edge
    -0.54
     Decay
    -0.54
     tongues
    -0.53
    POSITIVE LOGITS
    tical
    1.56
    tics
    1.52
    abouts
    1.52
    tic
    1.27
    with
    0.90
    ground
    0.75
    ford
    0.75
    after
    0.75
    ina
    0.72
    here
    0.71
    Activations Density 2.247%

    No Known Activations