© 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. 12505
    Prev
    Next
    INDEX
    Explanations

    significant or impactful phrases and notable names

    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
    156
    +0.65
    8.1%
    111
    +0.65
    8.1%
    25
    +0.07
    0.8%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    156
    +0.65
    0.42
    205
    +0.65
    0.48
    40
    +0.07
    0.47
    Negative Logits
    ĥ½
    -4.01
    č↵
    -3.97
    <|outofrange|>
    -3.97
    <|padding|>
    -3.97
    ↵↵
    -3.97
    <|outofrange|>
    -3.97
     ↵
    -3.97
    ↵
    -3.97
    ↵↵
    -3.97
    -3.97
    POSITIVE LOGITS
    rox
    1.04
    rosis
    1.00
     alone
    1.00
    olia
    0.99
    roscopic
    0.98
    ensive
    0.96
    wers
    0.91
     um
    0.91
    lez
    0.91
    frastructure
    0.90
    Activations Density 1.425%

    No Known Activations