© 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 EvalsExportsinterp-engineNEWAPI 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. 31261
    Prev
    Next
    INDEX
    Explanations

    structured programming constructs, particularly those related to function definitions and loops

    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
    320
    +0.14
    0.8%
    369
    +0.13
    0.7%
    51
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    51
    +0.14
    0.03
    188
    +0.13
    0.02
    320
    +0.12
    0.03
    Negative Logits
    **.
    -2.03
    .**
    -1.88
    **,
    -1.88
    **:
    -1.87
    .](
    -1.84
    .",
    -1.84
    :**
    -1.82
    icals
    -1.80
    "}\].
    -1.75
    ”.
    -1.68
    POSITIVE LOGITS
    ŀ
    2.13
    Ń
    1.71
    ·¸
    1.70
     keyword
    1.68
    ²
    1.67
    ĭ
    1.63
     extern
    1.60
    µ
    1.56
    ¾
    1.54
    uncture
    1.54
    Activations Density 0.180%

    No Known Activations