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. 11603
    Prev
    Next
    INDEX
    Explanations

    terms and phrases associated with medical and technical 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/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
    235
    +0.13
    0.8%
    174
    +0.13
    0.7%
    237
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    313
    +0.13
    0.05
    174
    +0.13
    0.13
    237
    +0.11
    0.12
    Negative Logits
     promptly
    -1.48
     matt
    -1.47
     mains
    -1.43
     Instr
    -1.40
    rame
    -1.36
     magazines
    -1.36
     soon
    -1.33
    blogger
    -1.32
     afterwards
    -1.31
     IU
    -1.31
    POSITIVE LOGITS
    aud
    1.58
    ',
    1.49
    uther
    1.49
    iment
    1.43
    ytes
    1.42
    ULAR
    1.38
    alth
    1.35
    **
    1.34
    erculosis
    1.34
    \'
    1.34
    Activations Density 4.655%

    No Known Activations