© 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. 4-RES-SM
    11. 10437
    Prev
    Next
    INDEX
    Explanations

    references to governmental or organizational changes and decisions

    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/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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    306
    +0.12
    0.7%
    183
    +0.11
    0.6%
    137
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    489
    +0.12
    0.03
    183
    +0.11
    0.04
    137
    +0.11
    0.04
    Negative Logits
    wide
    -1.64
    ahead
    -1.62
    ually
    -1.60
    """
    -1.57
     ourselves
    -1.55
    wegian
    -1.49
    atively
    -1.49
     oneself
    -1.48
     wide
    -1.48
     something
    -1.48
    POSITIVE LOGITS
     deux
    2.44
     même
    2.43
     prés
    2.23
     trois
    2.21
     rés
    2.20
     jour
    2.19
     nombre
    2.13
     dés
    2.09
     géné
    2.08
     représ
    2.06
    Activations Density 0.162%

    No Known Activations