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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & 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. 343
    Prev
    Next
    INDEX
    Explanations

    specific types of adjectives related to scenarios or events

    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
    362
    +0.17
    0.8%
    669
    +0.11
    0.5%
    408
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    362
    +0.17
    0.15
    669
    +0.11
    0.13
    408
    +0.09
    0.15
    Negative Logits
    agree
    -0.77
    uther
    -0.76
    mbudsman
    -0.76
    ploma
    -0.75
    ajor
    -0.75
    arthed
    -0.74
    mbuds
    -0.74
    á
    -0.73
    hers
    -0.73
    OIL
    -0.72
    POSITIVE LOGITS
    ly
    1.36
    ness
    1.36
    nesses
    1.07
     nature
    1.05
    est
    0.96
     sounding
    0.91
     tale
    0.88
    NESS
    0.85
     beginnings
    0.83
     exterior
    0.83
    Activations Density 1.987%

    No Known Activations