© 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. 1-RES-SM
    11. 25546
    Prev
    Next
    INDEX
    Explanations

    references to important subjects or concepts in the text

    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/1-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.1.hook_resid_post
    Hook Layer
    1
    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%
    98
    +0.20
    1.3%
    10
    +0.17
    1.1%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    10
    +0.24
    0.22
    98
    +0.20
    0.04
    291
    +0.17
    0.15
    Negative Logits
    ivat
    -1.58
    orf
    -1.54
     consecutive
    -1.45
     which
    -1.43
     contender
    -1.41
    cule
    -1.39
     Zel
    -1.38
     iteration
    -1.38
     seiz
    -1.36
     retros
    -1.35
    POSITIVE LOGITS
    ĨĴ
    2.60
    Ļª
    2.42
    Ĺ
    1.97
    ories
    1.93
    rapy
    1.93
    ·
    1.89
    opes
    1.87
    ses
    1.83
    rapeutics
    1.81
    ķ
    1.80
    Activations Density 2.307%

    No Known Activations