© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 1-RES-SM
    11. 29251
    Prev
    Next
    INDEX
    Explanations

    structured data in a specific format, likely code or configuration files

    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/1-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.1.hook_resid_post
    Hook Layer
    1
    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
    419
    +0.12
    0.7%
    76
    +0.11
    0.6%
    240
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    17
    +0.12
    0.04
    240
    +0.11
    0.01
    53
    +0.10
    0.06
    Negative Logits
    £
    -2.17
    ł
    -1.90
    ¾
    -1.89
    Ĺ
    -1.86
    ¿
    -1.84
    Ľ
    -1.77
    ĨĴ
    -1.73
    ī
    -1.71
    ĵ
    -1.69
    §
    -1.67
    POSITIVE LOGITS
     Kingdom
    1.55
     precious
    1.49
     voted
    1.45
     heard
    1.45
     compiled
    1.44
     Commonwealth
    1.43
     listened
    1.43
     testify
    1.41
     kingdom
    1.41
     Thames
    1.38
    Activations Density 0.884%

    No Known Activations