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 49152
    5. 8-RES_FS49152-JB
    6. 12051
    Prev
    Next
    INDEX
    Explanations

    instances of the word "mess" and related expressions indicating disorder or chaos

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    377
    +0.12
    0.6%
    526
    +0.11
    0.5%
    459
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    377
    +0.12
    0.01
    291
    +0.11
    0.01
    40
    +0.10
    0.01
    Negative Logits
    undai
    -0.83
     livest
    -0.79
    eele
    -0.74
    vation
    -0.66
     obser
    -0.65
    cember
    -0.64
    ©¶æ
    -0.61
     Pione
    -0.60
    éĸ
    -0.60
    cki
    -0.60
    POSITIVE LOGITS
    engers
    1.54
    iah
    1.21
    iness
    0.96
    enger
    0.95
    INESS
    0.93
    aging
    0.92
    aged
    0.86
    alos
    0.86
    es
    0.80
    ages
    0.80
    Activations Density 0.015%

    No Known Activations