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. 43301
    Prev
    Next
    INDEX
    Explanations

    actions related to falling or descent

    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
    335
    +0.12
    0.5%
    419
    +0.11
    0.5%
    268
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    335
    +0.12
    0.02
    9
    +0.11
    0.02
    469
    +0.10
    0.02
    Negative Logits
    iliate
    -0.74
    ²¾
    -0.69
     Remastered
    -0.68
     Gutenberg
    -0.68
    ansion
    -0.68
    zee
    -0.64
    adal
    -0.63
    iling
    -0.63
    redo
    -0.62
    Write
    -0.61
    POSITIVE LOGITS
     debris
    0.99
     asleep
    0.98
     bould
    0.92
     onto
    0.91
     overboard
    0.87
     ceilings
    0.85
    planes
    0.82
     cliffs
    0.82
     into
    0.82
     limp
    0.80
    Activations Density 0.036%

    No Known Activations