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. 29497
    Prev
    Next
    INDEX
    Explanations

    references to community-related events and interactions

    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
    288
    +0.17
    1.0%
    104
    +0.14
    0.8%
    118
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.17
    0.09
    145
    +0.14
    0.10
    248
    +0.13
    0.07
    Negative Logits
     harmless
    -1.70
    ([]
    -1.61
    ulative
    -1.53
     himself
    -1.52
     habits
    -1.50
    (['
    -1.49
     liar
    -1.48
     myself
    -1.47
     yourselves
    -1.45
     into
    -1.44
    POSITIVE LOGITS
     peak
    1.83
     apex
    1.73
     intersections
    1.72
     momento
    1.71
     intersection
    1.63
     periphery
    1.62
     center
    1.62
     terminus
    1.61
     Plaza
    1.61
     expense
    1.53
    Activations Density 1.287%

    No Known Activations