© 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. 12794
    Prev
    Next
    INDEX
    Explanations

    negative numerical values in mathematical expressions

    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
    125
    +0.12
    0.7%
    369
    +0.12
    0.7%
    358
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    506
    +0.12
    0.02
    125
    +0.12
    0.01
    358
    +0.11
    0.02
    Negative Logits
    acters
    -1.58
     successfully
    -1.53
    ↵
    -1.53
    abases
    -1.48
    ature
    -1.46
    atures
    -1.43
    ations
    -1.40
     conclusions
    -1.38
    andom
    -1.38
     itself
    -1.38
    POSITIVE LOGITS
    ¦
    3.47
    ĺ
    3.40
    ĥ½
    3.32
    Ħ
    3.26
    Ĵ
    3.15
    IJ
    3.07
    İ
    3.06
    ↵
    3.03
    3.03
    <|outofrange|>
    3.03
    Activations Density 0.031%

    No Known Activations