© 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. 27431
    Prev
    Next
    INDEX
    Explanations

    terms related to safeguarding and security

    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
    394
    +0.18
    1.0%
    277
    +0.13
    0.7%
    217
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    394
    +0.18
    0.05
    277
    +0.13
    0.04
    217
    +0.13
    0.04
    Negative Logits
    ĨĴ
    -2.21
    ¹
    -1.95
    acht
    -1.65
    ples
    -1.62
    ple
    -1.62
    ±
    -1.60
     startled
    -1.57
    enron
    -1.53
     pleas
    -1.51
     delighted
    -1.51
    POSITIVE LOGITS
     against
    2.20
     mechanisms
    2.09
    guards
    1.75
    ously
    1.72
     mechanism
    1.70
    ateral
    1.66
    iveness
    1.64
     weaknesses
    1.57
     protections
    1.56
     Against
    1.55
    Activations Density 0.300%

    No Known Activations