© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    3. Pythia-70M-Deduped
    4. Residuals
    5. 2-RES-SM
    6. 12383
    Prev
    Next
    INDEX
    Explanations

    references to statistical data or quantitative information

    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
    369
    +0.16
    0.9%
    496
    +0.13
    0.7%
    510
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    510
    +0.16
    0.13
    284
    +0.13
    0.11
    152
    +0.13
    0.08
    Negative Logits
    inux
    -1.75
    sible
    -1.54
    ctica
    -1.48
    gel
    -1.46
    imiento
    -1.45
     avatar
    -1.38
    issance
    -1.37
    yrinth
    -1.36
     Ãħ
    -1.34
    "?
    -1.34
    POSITIVE LOGITS
     authors
    1.95
     investigators
    1.86
     researchers
    1.78
     authorities
    1.74
     commentators
    1.72
    áĢº
    1.69
     resear
    1.66
     respondents
    1.66
    ographers
    1.60
    Plaintiffs
    1.58
    Activations Density 2.032%

    No Known Activations