© 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
    2. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    3. Pythia-70M-Deduped
    4. Residuals
    5. 2-RES-SM
    6. 31747
    Prev
    Next
    INDEX
    Explanations

    phrases indicating conditions and requirements regarding contracts or agreements

    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
    156
    +0.24
    1.5%
    235
    +0.21
    1.3%
    323
    +0.15
    0.9%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    156
    +0.24
    0.76
    111
    +0.21
    0.79
    71
    +0.15
    0.71
    Negative Logits
    vare
    -1.51
     why
    -1.48
    opolys
    -1.41
    asa
    -1.41
    vez
    -1.41
    arxiv
    -1.36
    onas
    -1.35
     pathogenesis
    -1.34
    return
    -1.33
    ameth
    -1.33
    POSITIVE LOGITS
    ="{
    1.47
     officer
    1.33
     observer
    1.31
     Olymp
    1.28
    holders
    1.26
     officers
    1.26
     England
    1.26
     martial
    1.25
    listed
    1.25
     holder
    1.23
    Activations Density 4.611%

    No Known Activations