© 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. 10624
    Prev
    Next
    INDEX
    Explanations

    mentions of the name "James" 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/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
    +0.20
    1.1%
    410
    +0.14
    0.8%
    172
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    410
    +0.20
    0.03
    369
    +0.14
    0.02
    193
    +0.12
    0.02
    Negative Logits
     otherwise
    -2.17
    dered
    -1.79
    matically
    -1.71
    ten
    -1.66
    ASE
    -1.65
    izer
    -1.65
    ties
    -1.64
    ational
    -1.63
    ths
    -1.63
    ted
    -1.60
    POSITIVE LOGITS
     Bond
    2.00
     Comey
    1.93
     Clerk
    1.86
    andro
    1.69
    boro
    1.65
    on
    1.64
     Madison
    1.62
     Baldwin
    1.62
     Patterson
    1.62
    ville
    1.57
    Activations Density 0.126%

    No Known Activations