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 24576
    5. 8-RES_FS24576-JB
    6. 23494
    Prev
    Next
    INDEX
    Explanations

    instances of the phrase "shortly after."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    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
    210
    +0.09
    0.4%
    183
    +0.09
    0.4%
    172
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    172
    +0.09
    0.03
    210
    +0.09
    0.03
    98
    +0.09
    0.03
    Negative Logits
     unfocusedRange
    -0.80
    aez
    -0.78
    avis
    -0.73
    enberg
    -0.73
    ometimes
    -0.69
    owed
    -0.67
    acles
    -0.67
    adj
    -0.66
    amount
    -0.66
    ahime
    -0.65
    POSITIVE LOGITS
    noon
    1.20
     midnight
    1.08
     noon
    1.02
     arriving
    0.94
     Thanksgiving
    0.88
     sunset
    0.84
     sunrise
    0.82
     completion
    0.82
    words
    0.81
     inception
    0.80
    Activations Density 0.058%

    No Known Activations