© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 4-RES-SM
    11. 8775
    Prev
    Next
    INDEX
    Explanations

    specific identifiers and terminology related to the healthcare and medical field

    oai_token-act-pair · gpt-4o-mini
    Triggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    ctigges/pythia-70m-deduped__res-sm_processed/4-res-sm
    Prompts (Dashboard)
    32,768 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.4.hook_resid_post
    Hook Layer
    4
    Architecture
    standard
    Context Size
    128
    Dataset
    EleutherAI/the_pile_deduplicated
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Neuron Alignment
    Index
    Value
    % of L₁
    156
    +0.21
    1.2%
    186
    +0.15
    0.9%
    25
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    440
    +0.21
    0.02
    186
    +0.15
    0.09
    421
    +0.13
    0.11
    Negative Logits
    ij
    -3.74
    ¸
    -3.53
    ¯
    -3.48
    ±
    -3.35
    ĥ
    -3.30
    ģ
    -3.29
    Ļª
    -3.27
    ĨĴ
    -3.25
    ·
    -3.18
    ↵
    -3.18
    POSITIVE LOGITS
     policies
    2.02
    's
    2.00
    âĦ¢
    1.91
    ELY
    1.89
     fans
    1.87
     membership
    1.87
     supporters
    1.87
     enthusiasts
    1.84
    ®
    1.81
     counterparts
    1.78
    Activations Density 2.175%

    No Known Activations