© 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 EvalsExportsinterp-engineNEWAPI 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. 16763
    Prev
    Next
    INDEX
    Explanations

    technical terms related to technology and finance

    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
    373
    +0.12
    0.6%
    331
    +0.10
    0.5%
    309
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    255
    +0.12
    0.04
    164
    +0.10
    0.04
    546
    +0.09
    0.03
    Negative Logits
    selage
    -0.65
    izens
    -0.65
     Urug
    -0.65
     Yon
    -0.65
     Sung
    -0.62
     prince
    -0.62
    oka
    -0.61
     loyal
    -0.61
    swick
    -0.61
    âĶĢâĶĢ
    -0.61
    POSITIVE LOGITS
     whereby
    1.21
     causation
    0.94
     infiltration
    0.89
     Techniques
    0.88
    efficiency
    0.88
     aka
    0.88
     stunts
    0.85
     occurs
    0.84
     involves
    0.82
     techniques
    0.82
    Activations Density 0.432%

    No Known Activations