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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlogSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 3072
    5. 8-RES_FS3072-JB
    6. 1180
    Prev
    Next
    INDEX
    Explanations

    technical information related to software development and protocols

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs3072-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    3,072
    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
    311
    +0.13
    0.6%
    620
    +0.12
    0.5%
    514
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    311
    +0.13
    0.02
    326
    +0.12
    0.04
    514
    +0.10
    0.03
    Negative Logits
    ItemImage
    -0.69
    ercise
    -0.60
    perty
    -0.59
     tabl
    -0.58
     oath
    -0.57
    erity
    -0.57
    creen
    -0.56
     cold
    -0.56
    akov
    -0.56
    ages
    -0.55
    POSITIVE LOGITS
    xff
    1.33
    xd
    1.03
    xa
    0.99
    xc
    0.96
    644
    0.92
    xb
    0.92
    xe
    0.89
    x
    0.89
    755
    0.88
    603
    0.84
    Activations Density 0.470%

    No Known Activations