© 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. 4-RES-SM
    11. 11590
    Prev
    Next
    INDEX
    Explanations

    XML-like markup or code structure

    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/4-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.4.hook_resid_post
    Hook Layer
    4
    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
    21
    +0.13
    0.8%
    6
    +0.10
    0.6%
    375
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    375
    +0.13
    0.04
    461
    +0.10
    0.04
    103
    +0.10
    0.04
    Negative Logits
    INGTON
    -1.83
    pex
    -1.82
     themselves
    -1.46
     itself
    -1.42
    atrix
    -1.41
    soever
    -1.38
    inson
    -1.32
    hetti
    -1.32
    Collins
    -1.30
    etin
    -1.30
    POSITIVE LOGITS
    ģ
    2.35
    summary
    2.10
    İ
    2.08
     =>
    2.04
    Ķ
    1.97
    ŀ
    1.96
    ¥
    1.94
    ĵ
    1.89
    inherit
    1.83
    ĭ
    1.81
    Activations Density 0.196%

    No Known Activations