Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    3. Pythia-70M-Deduped
    4. Residuals
    5. 2-RES-SM
    6. 29414
    Prev
    Next
    INDEX
    Explanations

    references to the Internet and its associated technologies

    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/2-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.2.hook_resid_post
    Hook Layer
    2
    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
    +0.21
    1.2%
    451
    +0.12
    0.7%
    460
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    451
    +0.21
    0.01
    460
    +0.12
    0.02
    173
    +0.11
    0.01
    Negative Logits
     drastic
    -1.88
     uterus
    -1.72
    Į
    -1.71
     solitary
    -1.66
    unts
    -1.64
     tear
    -1.58
     same
    -1.54
    aucoup
    -1.53
     certain
    -1.52
     resign
    -1.51
    POSITIVE LOGITS
     Explorer
    1.99
    borne
    1.91
    works
    1.85
     Access
    1.79
    Gate
    1.73
    Link
    1.73
     Resources
    1.61
    burst
    1.59
    sphere
    1.57
     Archive
    1.55
    Activations Density 0.090%

    No Known Activations