© 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. 747
    Prev
    Next
    INDEX
    Explanations

    words related to communication and discussion in a professional or interpersonal setting

    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
    493
    +0.10
    0.4%
    679
    +0.10
    0.4%
    380
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    70
    +0.10
    0.12
    195
    +0.10
    0.05
    679
    +0.10
    0.14
    Negative Logits
    hews
    -0.68
    aredevil
    -0.68
    \/\/
    -0.67
     Stores
    -0.67
     Soldiers
    -0.67
    arte
    -0.66
    uilt
    -0.65
     landfall
    -0.61
    rypt
    -0.60
    xtap
    -0.57
    POSITIVE LOGITS
     louder
    1.04
     spoken
    1.02
     aloud
    0.99
     voices
    0.95
     transcripts
    0.94
     about
    0.92
     loudly
    0.92
    voice
    0.85
     talk
    0.84
     fluent
    0.84
    Activations Density 4.401%

    No Known Activations