© 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 EvalsExportsAPI 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. 9740
    Prev
    Next
    INDEX
    Explanations

    phrases indicating relationships, particularly conjunctions and other connective words

    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
    478
    +0.18
    1.0%
    258
    +0.14
    0.8%
    314
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    209
    +0.18
    0.11
    258
    +0.14
    0.12
    494
    +0.11
    0.06
    Negative Logits
    Īĺ
    -3.25
    ģ
    -2.65
    Ģ
    -2.58
    »¿
    -2.48
    Ł
    -2.41
    ij
    -2.39
    ·
    -2.33
    ĥ
    -2.33
    ¸
    -2.27
    ¿½
    -2.24
    POSITIVE LOGITS
     consequent
    1.66
    rogens
    1.61
     others
    1.59
     yes
    1.55
     vice
    1.52
    alus
    1.50
     willingness
    1.50
    rei
    1.49
    iamo
    1.48
     other
    1.48
    Activations Density 1.133%

    No Known Activations