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. 414
    Prev
    Next
    INDEX
    Explanations

    exclamations or expressions of strong emotion

    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
    480
    +0.39
    2.1%
    447
    +0.30
    1.6%
    87
    +0.18
    0.9%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    266
    +0.39
    0.20
    480
    +0.30
    0.06
    119
    +0.18
    0.10
    Negative Logits
    sequently
    -0.66
     subsequent
    -0.60
    Mobil
    -0.59
    ãĥ¼ãĥĨ
    -0.58
    luster
    -0.57
     helicop
    -0.55
     Azerb
    -0.55
     subsequently
    -0.55
     eventual
    -0.53
     redes
    -0.53
    POSITIVE LOGITS
     gonna
    0.90
     shit
    0.87
    !!!!
    0.87
     fucked
    0.86
     dont
    0.86
     fuck
    0.84
     ain
    0.83
     fuckin
    0.83
     fucking
    0.83
     doesnt
    0.83
    Activations Density 3.973%

    No Known Activations