© 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 768
    9. 8-RES_FS768-JB
    11. 727
    Prev
    Next
    INDEX
    Explanations

    personal possessive pronouns and possessive determiners

    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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    362
    +0.19
    0.9%
    385
    +0.15
    0.7%
    608
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    362
    +0.19
    0.17
    385
    +0.15
    0.14
    167
    +0.11
    0.05
    Negative Logits
    avis
    -0.78
    wrap
    -0.78
    olkien
    -0.77
    etz
    -0.74
    yang
    -0.72
    ]'
    -0.70
    ciation
    -0.69
     Goes
    -0.69
    annot
    -0.69
    udi
    -0.69
    POSITIVE LOGITS
     own
    1.63
     favorite
    1.41
     favourite
    1.33
     beloved
    1.18
    selves
    1.14
     opponent
    1.14
     grandmother
    1.12
     ancestors
    1.12
     daughter
    1.11
     hometown
    1.11
    Activations Density 1.724%

    No Known Activations