© 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. 30795
    Prev
    Next
    INDEX
    Explanations

    syntactical elements in programming languages, particularly involving variables and control flow logic

    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
    47
    +0.14
    0.8%
    482
    +0.13
    0.7%
    267
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    47
    +0.14
    0.07
    29
    +0.13
    0.03
    22
    +0.12
    0.04
    Negative Logits
    STM
    -2.02
     exploits
    -1.51
    Ħ
    -1.50
    tools
    -1.49
    au
    -1.45
    ulose
    -1.45
    Copyright
    -1.44
    /#
    -1.40
    \]](
    -1.37
    MAN
    -1.36
    POSITIVE LOGITS
     extremes
    1.98
     majority
    1.55
    ftime
    1.50
     wait
    1.48
     cancel
    1.48
    NdEx
    1.47
     ordinal
    1.45
     monoton
    1.43
     offset
    1.37
     hurry
    1.37
    Activations Density 0.473%

    No Known Activations