© 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. 31794
    Prev
    Next
    INDEX
    Explanations

    dates, particularly those in June

    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
    43
    +0.21
    1.2%
    38
    +0.13
    0.7%
    82
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    43
    +0.21
    0.03
    82
    +0.13
    0.02
    319
    +0.11
    0.02
    Negative Logits
     quo
    -1.72
    inous
    -1.68
    ´
    -1.65
    catch
    -1.62
    nip
    -1.58
    erver
    -1.56
    versus
    -1.51
    herent
    -1.51
    ism
    -1.46
    hip
    -1.46
    POSITIVE LOGITS
    ruary
    1.90
    osph
    1.72
    vier
    1.70
    bourne
    1.70
    rile
    1.63
     onwards
    1.61
    vue
    1.58
    forth
    1.58
    uary
    1.55
    au
    1.52
    Activations Density 0.059%

    No Known Activations