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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 1536
    5. 8-RES_FS1536-JB
    6. 370
    Prev
    Next
    INDEX
    Explanations

    phrases related to being the "world's largest" or "world's first" in various contexts

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs1536-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    1,536
    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
    203
    +0.10
    0.4%
    79
    +0.09
    0.4%
    558
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    67
    +0.10
    0.07
    558
    +0.09
    0.07
    203
    +0.09
    0.06
    Negative Logits
    ippi
    -0.96
    veyard
    -0.82
    »Ĵ
    -0.79
    ibility
    -0.76
    imon
    -0.74
    iple
    -0.73
    edin
    -0.72
     Mn
    -0.72
    sbm
    -0.72
    ifer
    -0.71
    POSITIVE LOGITS
    wide
    1.05
    liness
    0.91
     premiere
    0.87
    building
    0.85
     Wide
    0.84
    wake
    0.84
    Wide
    0.79
     renowned
    0.78
     championships
    0.76
     religions
    0.76
    Activations Density 1.064%

    No Known Activations