© 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. 11126
    Prev
    Next
    INDEX
    Explanations

    references to individual pieces of items or components

    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
    502
    +0.14
    0.8%
    144
    +0.13
    0.7%
    174
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    144
    +0.14
    0.02
    502
    +0.13
    0.02
    249
    +0.12
    0.01
    Negative Logits
    ĥ
    -3.24
    ľĵ
    -3.23
    Ń
    -3.10
    ĥ½
    -3.02
    ĨĴ
    -3.00
    Ĥ
    -2.87
    ĺ
    -2.87
    ·
    -2.84
    ľ
    -2.83
    ĸ´
    -2.82
    POSITIVE LOGITS
    meal
    3.97
    wise
    2.39
    board
    1.98
    nam
    1.93
    heit
    1.90
    boards
    1.83
    nel
    1.81
     woven
    1.81
    pieces
    1.80
    icles
    1.76
    Activations Density 0.186%

    No Known Activations