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

    Neuronpedia

    Jacobian LensNEW
    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. 17376
    Prev
    Next
    INDEX
    Explanations

    languages and language-related terms from various cultures

    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
    218
    +0.13
    0.6%
    373
    +0.11
    0.5%
    61
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    103
    +0.13
    0.03
    406
    +0.11
    0.03
    107
    +0.11
    0.02
    Negative Logits
     Nightmares
    -0.67
    utics
    -0.66
     Catalyst
    -0.61
    ilogy
    -0.61
    ulators
    -0.60
     Faces
    -0.60
    istries
    -0.59
     prompts
    -0.59
    =-=-=-=-
    -0.59
     Parish
    -0.58
    POSITIVE LOGITS
    çĶ°
    1.12
    Ñģ
    1.07
    к
    1.05
    Ð
    1.03
    д
    1.02
    ÑĤ
    1.01
    ÑĢ
    1.00
    л
    0.99
    λ
    0.97
    é»Ĵ
    0.94
    Activations Density 0.041%

    No Known Activations