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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 24576
    9. 8-RES_FS24576-JB
    11. 2164
    Prev
    Next
    INDEX
    Explanations

    mentions of legal actions or charges against individuals

    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
    488
    +0.13
    0.6%
    408
    +0.12
    0.5%
    102
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    488
    +0.13
    0.03
    59
    +0.12
    0.02
    104
    +0.11
    0.02
    Negative Logits
     ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³
    -0.91
    aucuses
    -0.85
    soDeliveryDate
    -0.82
     helicop
    -0.81
    ãĥ¼ãĥĨãĤ£
    -0.80
    ãĤĬ
    -0.77
    ãĤ¼ãĤ¦ãĤ¹
    -0.76
    ahime
    -0.74
     teasp
    -0.74
    姫
    -0.74
    POSITIVE LOGITS
     him
    0.98
     defendants
    0.89
     alleged
    0.84
     whistleblowers
    0.83
     anyone
    0.81
     innocent
    0.79
     anybody
    0.77
     individuals
    0.77
     fellow
    0.76
     wrongdoing
    0.76
    Activations Density 0.059%

    No Known Activations