© 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. 12800
    Prev
    Next
    INDEX
    Explanations

    mentions of legal proceedings or related legal terminology

    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
    21
    +0.36
    2.9%
    306
    +0.31
    2.5%
    307
    +0.21
    1.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    450
    +0.36
    0.16
    21
    +0.31
    0.14
    306
    +0.21
    0.16
    Negative Logits
     bases
    -1.39
     base
    -1.31
     formation
    -1.17
     hold
    -1.14
     utility
    -1.13
     codes
    -1.11
     aim
    -1.10
     system
    -1.08
     done
    -1.08
     rewards
    -1.07
    POSITIVE LOGITS
     »
    1.28
    1451450014514500
    1.27
    "</
    1.25
     Atty
    1.24
    into
    1.23
    àº
    1.20
    Reuters
    1.19
    Decided
    1.17
    undle
    1.16
    Asked
    1.16
    Activations Density 0.695%

    No Known Activations