© 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. 29622
    Prev
    Next
    INDEX
    Explanations

    occurrences of the word "despite" or phrases indicating contrast or adversity

    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
    23
    +0.26
    1.5%
    376
    +0.18
    1.0%
    202
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    202
    +0.26
    0.03
    46
    +0.18
    0.02
    316
    +0.12
    0.02
    Negative Logits
    anse
    -1.85
    aurus
    -1.78
     assists
    -1.67
    inos
    -1.65
    onic
    -1.61
    ossier
    -1.58
    #:
    -1.58
    leen
    -1.57
    ricular
    -1.51
     vs
    -1.47
    POSITIVE LOGITS
     previous
    1.80
     sorts
    1.74
     warnings
    1.72
     slight
    1.68
    noon
    1.60
     hers
    1.60
    tons
    1.58
     luck
    1.58
     slightest
    1.55
     careful
    1.50
    Activations Density 0.128%

    No Known Activations