© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    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. 10785
    Prev
    Next
    INDEX
    Explanations

    references to air-related topics

    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
    156
    +0.17
    1.0%
    443
    +0.15
    0.9%
    164
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    443
    +0.17
    0.03
    304
    +0.15
    0.03
    113
    +0.13
    0.03
    Negative Logits
     married
    -1.64
    thood
    -1.54
     hers
    -1.53
     grandchildren
    -1.53
    hered
    -1.51
     autoimmune
    -1.45
     honour
    -1.45
     appointment
    -1.44
    nae
    -1.43
     systematic
    -1.41
    POSITIVE LOGITS
    Ĵ
    2.64
    ħ
    2.27
    ģ
    2.12
    waves
    2.11
    ĺ
    2.07
    ®
    2.05
    «
    2.01
    ľ
    2.01
    ŀ
    2.00
    ports
    2.00
    Activations Density 0.103%

    No Known Activations