© 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. 31215
    Prev
    Next
    INDEX
    Explanations

    mentions of additional or extra quantities or features

    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
    129
    +0.11
    0.6%
    323
    +0.10
    0.6%
    86
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    430
    +0.11
    0.02
    148
    +0.10
    0.00
    179
    +0.10
    0.02
    Negative Logits
    sis
    -1.84
    quez
    -1.75
    dez
    -1.73
    ollah
    -1.68
    aus
    -1.68
    uca
    -1.63
    ockey
    -1.60
    onson
    -1.57
    oya
    -1.53
    ken
    -1.52
    POSITIVE LOGITS
    ordinary
    2.27
    judicial
    2.14
     layers
    1.88
    isting
    1.80
     virgin
    1.77
    Seconds
    1.73
     than
    1.70
    ctions
    1.65
     protections
    1.58
     layer
    1.58
    Activations Density 0.187%

    No Known Activations