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. 29595
    Prev
    Next
    INDEX
    Explanations

    the presence of periods or sentence endings in the text

    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
    111
    +0.74
    6.5%
    156
    +0.28
    2.5%
    258
    +0.10
    0.9%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    111
    +0.74
    0.73
    156
    +0.28
    0.62
    71
    +0.10
    0.60
    Negative Logits
    ľĵ
    -3.72
    Į
    -3.26
    ¿½
    -3.16
    ¿
    -3.06
    ¦
    -3.00
    »
    -2.95
    ¹
    -2.94
    ij
    -2.93
    ®
    -2.91
    Īĺ
    -2.87
    POSITIVE LOGITS
     fundament
    1.38
     polÃŃ
    1.34
     hype
    1.29
    actory
    1.27
     parametric
    1.24
     patent
    1.23
     patents
    1.21
     UE
    1.20
     antitrust
    1.20
     operating
    1.18
    Activations Density 0.809%

    No Known Activations