© 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. 32003
    Prev
    Next
    INDEX
    Explanations

    causal relationships and dependencies in various contexts

    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
    450
    +0.15
    0.9%
    221
    +0.12
    0.6%
    191
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    450
    +0.15
    0.03
    215
    +0.12
    0.03
    191
    +0.11
    0.03
    Negative Logits
    ĨĴ
    -1.77
    Ļª
    -1.75
    mouth
    -1.57
     Election
    -1.49
    blogger
    -1.45
    ¢
    -1.45
    sburg
    -1.43
    ĭ
    -1.40
    "))
    -1.34
     polls
    -1.32
    POSITIVE LOGITS
     regard
    1.66
    iment
    1.65
    iments
    1.60
    ences
    1.49
     concer
    1.45
    ño
    1.45
    initis
    1.44
    eps
    1.44
     deference
    1.42
    idden
    1.42
    Activations Density 0.106%

    No Known Activations