© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 12288
    9. 8-RES_FS12288-JB
    11. 1363
    Prev
    Next
    INDEX
    Explanations

    first person pronouns in sentences talking about actions or experiences

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    502
    +0.12
    0.5%
    638
    +0.11
    0.5%
    631
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    502
    +0.12
    0.08
    566
    +0.11
    0.06
    551
    +0.10
    0.06
    Negative Logits
    noon
    -0.68
    mon
    -0.67
    mony
    -0.61
    aston
    -0.60
     Nationwide
    -0.59
    whel
    -0.59
    church
    -0.59
    bender
    -0.58
    mart
    -0.57
    termination
    -0.56
    POSITIVE LOGITS
     learnt
    0.95
     wrought
    0.84
     fuss
    0.84
     accomplished
    0.84
     learned
    0.84
    've
    0.80
     termed
    0.80
     perceive
    0.79
     glean
    0.77
     wanted
    0.76
    Activations Density 0.231%

    No Known Activations