© 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. 28190
    Prev
    Next
    INDEX
    Explanations

    references to chemical concentrations or levels in various contexts

    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
    156
    +0.16
    0.9%
    498
    +0.12
    0.6%
    164
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    424
    +0.16
    0.01
    498
    +0.12
    0.01
    232
    +0.11
    0.01
    Negative Logits
    ĻĤ
    -1.94
    ½
    -1.87
    nered
    -1.75
    INGTON
    -1.60
    nae
    -1.55
    ĵ
    -1.52
    ľĵ
    -1.51
     nephe
    -1.51
    ¾
    -1.49
     blessed
    -1.47
    POSITIVE LOGITS
     ranges
    2.16
    zone
    1.78
     range
    1.76
    frame
    1.72
     tolerance
    1.70
    buffer
    1.68
    limits
    1.65
     amounts
    1.65
     varies
    1.61
    gebra
    1.60
    Activations Density 0.046%

    No Known Activations