© 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. 10014
    Prev
    Next
    INDEX
    Explanations

    specific technical terms and data related to programming or software specifications

    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
    210
    +0.13
    0.7%
    280
    +0.11
    0.6%
    198
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    210
    +0.13
    0.02
    399
    +0.11
    0.03
    39
    +0.11
    0.05
    Negative Logits
     violated
    -1.54
     abroad
    -1.49
    lessness
    -1.40
    arium
    -1.39
     demanding
    -1.39
    wort
    -1.33
     worth
    -1.31
     stolen
    -1.26
     Laboratories
    -1.26
     exist
    -1.24
    POSITIVE LOGITS
    Ļª
    3.00
    ¶
    2.95
    ģ
    2.84
    ¦
    2.80
    Īĺ
    2.79
    ı
    2.74
    ŀ
    2.71
    ĵ
    2.70
    ĻĤ
    2.68
    ĥ
    2.65
    Activations Density 4.615%

    No Known Activations