© 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. 31070
    Prev
    Next
    INDEX
    Explanations

    conjunctions and transitional words that introduce contrasting or conditional ideas

    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
    135
    +0.15
    0.8%
    465
    +0.12
    0.7%
    176
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    496
    +0.15
    0.06
    135
    +0.12
    0.05
    123
    +0.12
    0.05
    Negative Logits
    imeters
    -1.70
    IER
    -1.36
    erent
    -1.33
    ther
    -1.30
    cias
    -1.30
     wel
    -1.30
     —
    -1.28
    atur
    -1.28
     pneum
    -1.27
    aurus
    -1.25
    POSITIVE LOGITS
              ↵
    1.96
    č↵
    1.96
    ↵
    1.96
    ↵
    1.96
    ↵
    1.96
    <|outofrange|>
    1.96
    1.96
    <|outofrange|>
    1.96
    1.96
     ↵
    1.96
    Activations Density 0.276%

    No Known Activations