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. 685
    Prev
    Next
    INDEX
    Explanations

    references to educational qualifications and technical skills

    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
    730
    +0.11
    0.5%
    169
    +0.10
    0.5%
    68
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    14
    +0.11
    0.08
    55
    +0.10
    0.10
    5
    +0.10
    0.07
    Negative Logits
    terness
    -0.92
    icism
    -0.91
    matter
    -0.89
    dom
    -0.85
    cation
    -0.84
    oire
    -0.83
    ileaks
    -0.82
    ashi
    -0.79
    mentation
    -0.78
    antry
    -0.77
    POSITIVE LOGITS
    wolves
    0.99
     constants
    0.97
     examples
    0.97
     exceptions
    0.96
    types
    0.95
     snapshots
    0.89
     indications
    0.88
     embodiments
    0.88
     nods
    0.87
     reminders
    0.86
    Activations Density 1.736%

    No Known Activations