© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 4-RES-SM
    11. 11254
    Prev
    Next
    INDEX
    Explanations

    references to groundbreaking innovations or developments

    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/4-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.4.hook_resid_post
    Hook Layer
    4
    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
    164
    +0.12
    0.7%
    127
    +0.11
    0.6%
    233
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    164
    +0.12
    0.03
    233
    +0.11
    0.02
    195
    +0.11
    0.02
    Negative Logits
    TRODUCTION
    -1.62
    olic
    -1.59
    olta
    -1.52
    late
    -1.51
    choline
    -1.48
    cent
    -1.45
    isters
    -1.42
    ority
    -1.41
    ugin
    -1.40
    sembles
    -1.39
    POSITIVE LOGITS
    breaking
    2.77
    hog
    2.74
    ing
    2.23
    nuts
    2.21
    work
    2.17
    truth
    2.11
    worthy
    2.02
    nut
    2.02
     floor
    2.01
    fall
    1.99
    Activations Density 0.152%

    No Known Activations