Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlog/PodcastSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlog/PodcastGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 24576
    5. 8-RES_FS24576-JB
    6. 811
    Prev
    Next
    INDEX
    Explanations

    features/items of a mechanical object, particularly focusing on the components or parts of the object like coil springs, magazine extension, grip, hose, tube, screw-on cap, frame mounts, flat bar, pinch seam, rockers, fork, tubing, bracket, and spring rates

    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
    420
    +0.13
    0.6%
    447
    +0.12
    0.5%
    415
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    678
    +0.13
    0.06
    326
    +0.12
    0.05
    41
    +0.10
    0.05
    Negative Logits
    ãģĦ
    -0.77
    isine
    -0.70
    rians
    -0.67
    incarn
    -0.66
     Incarn
    -0.66
    ceans
    -0.66
    abama
    -0.66
    Politics
    -0.65
     visions
    -0.65
    ModLoader
    -0.64
    POSITIVE LOGITS
     protr
    1.12
     diameter
    1.08
     attachment
    1.06
     weld
    1.06
     attached
    1.05
     molded
    1.04
     fitted
    1.04
     connector
    1.03
     clamp
    1.00
     attaches
    0.95
    Activations Density 0.222%

    No Known Activations