© 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. 10576
    Prev
    Next
    INDEX
    Explanations

    versions and release information of software or libraries

    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
    177
    +0.14
    0.8%
    509
    +0.12
    0.7%
    193
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    509
    +0.14
    0.03
    193
    +0.12
    0.03
    500
    +0.11
    0.03
    Negative Logits
    pute
    -1.54
    ceptions
    -1.53
    mer
    -1.51
    amus
    -1.42
    governmental
    -1.42
    ancers
    -1.40
    shore
    -1.36
    ouses
    -1.36
    umni
    -1.35
    piece
    -1.30
    POSITIVE LOGITS
    ing
    2.38
    ed
    2.01
    ĭ
    1.91
     control
    1.85
     number
    1.81
    number
    1.80
     thereof
    1.78
    eda
    1.77
     numbers
    1.76
    ahead
    1.73
    Activations Density 0.174%

    No Known Activations