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. 46003
    Prev
    Next
    INDEX
    Explanations

    references to time, particularly the word "later" and its various usages in context

    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
    481
    +0.12
    0.5%
    132
    +0.11
    0.5%
    430
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    488
    +0.12
    0.02
    300
    +0.11
    0.02
    564
    +0.10
    0.02
    Negative Logits
     Scal
    -0.70
    hab
    -0.69
    lua
    -0.65
    THING
    -0.61
    rament
    -0.61
     SVG
    -0.61
    owship
    -0.58
     RAID
    -0.58
     NFC
    -0.58
    effic
    -0.57
    POSITIVE LOGITS
    aneously
    0.81
    forth
    0.80
     when
    0.76
    .</
    0.73
     than
    0.72
     regretted
    0.71
     onwards
    0.69
     succumbed
    0.68
    iday
    0.66
     succumb
    0.64
    Activations Density 0.017%

    No Known Activations