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

    Neuronpedia

    APISteerSAE EvalsBlog/PodcastNEWSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlog/PodcastGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 768
    5. 8-RES_FS768-JB
    6. 264
    Prev
    Next
    INDEX
    Explanations

    adverbs and adjectives that express doubt or certainty

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs768-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    768
    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
    •"make sure"
    "surely"/"no doubt"/"probably"
    •Make sure feature
    •Canonical Figure 13

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    8
    +0.11
    0.5%
    760
    +0.10
    0.5%
    231
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    8
    +0.11
    0.15
    718
    +0.10
    0.13
    596
    +0.10
    0.10
    Negative Logits
    uality
    -0.78
    idas
    -0.76
    irl
    -0.74
    ocaust
    -0.71
    iple
    -0.71
    ËĪ
    -0.71
    isi
    -0.71
    psey
    -0.71
    zanne
    -0.71
    ERAL
    -0.70
    POSITIVE LOGITS
     unsurprisingly
    0.79
     understatement
    0.72
     underest
    0.69
     someday
    0.68
     deserved
    0.67
     horr
    0.65
     SOME
    0.65
     Admir
    0.64
     qualifies
    0.64
     whoever
    0.62
    Activations Density 3.827%

    No Known Activations