© 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 24576
    9. 8-RES_FS24576-JB
    11. 9366
    Prev
    Next
    INDEX
    Explanations

    information related to event organization and participation instructions

    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
    520
    +0.09
    0.4%
    646
    +0.09
    0.4%
    31
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    162
    +0.09
    0.04
    520
    +0.09
    0.04
    302
    +0.09
    0.04
    Negative Logits
    âĹ¼
    -0.68
    æ©Ł
    -0.67
     Zimmer
    -0.65
    å®
    -0.64
    Redd
    -0.63
    ãĤ·ãĥ£
    -0.63
     Assy
    -0.62
    Jere
    -0.62
     folklore
    -0.62
    ãĥķãĤ¡
    -0.62
    POSITIVE LOGITS
     maximum
    0.99
     unlimited
    0.98
     prepaid
    0.92
    maximum
    0.92
     Maximum
    0.87
     increments
    0.85
    Maximum
    0.78
     servings
    0.77
     inclusive
    0.76
     occupancy
    0.76
    Activations Density 0.366%

    No Known Activations