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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 2-RES-SM
    11. 27173
    Prev
    Next
    INDEX
    Explanations

    terms and expressions related to technology industry metrics and support structures

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    ctigges/pythia-70m-deduped__res-sm_processed/2-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.2.hook_resid_post
    Hook Layer
    2
    Architecture
    standard
    Context Size
    128
    Dataset
    EleutherAI/the_pile_deduplicated
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    271
    +0.17
    1.0%
    293
    +0.13
    0.8%
    201
    +0.13
    0.8%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    201
    +0.17
    0.06
    248
    +0.13
    0.09
    293
    +0.13
    0.09
    Negative Logits
     basis
    -1.62
     schedule
    -1.55
    inks
    -1.45
     fertility
    -1.44
     autom
    -1.39
     cured
    -1.39
     dosage
    -1.33
     hers
    -1.31
    ake
    -1.30
     advantage
    -1.30
    POSITIVE LOGITS
     (@
    1.61
     whom
    1.40
    main
    1.37
     alike
    1.36
     convicted
    1.36
    JJ
    1.34
     mob
    1.33
    JB
    1.31
    mber
    1.31
     Stre
    1.30
    Activations Density 3.083%

    No Known Activations