© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    2. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    3. Pythia-70M-Deduped
    4. Residuals
    5. 2-RES-SM
    6. 29870
    Prev
    Next
    INDEX
    Explanations

    elements related to coding, configurations, and specific technical details about frameworks and libraries

    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
    231
    +0.18
    1.0%
    436
    +0.18
    1.0%
    419
    +0.17
    1.0%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    4
    +0.18
    0.13
    156
    +0.18
    0.14
    111
    +0.17
    0.12
    Negative Logits
    zyk
    -1.61
    ivities
    -1.53
     cytometry
    -1.48
     lent
    -1.42
    ondo
    -1.42
    ption
    -1.40
     habit
    -1.38
    =”
    -1.36
    ulsion
    -1.33
    agne
    -1.33
    POSITIVE LOGITS
     basis
    1.60
    done
    1.54
    shirts
    1.53
    rible
    1.32
    anked
    1.29
    anded
    1.25
     bases
    1.23
     symbol
    1.21
    buntu
    1.21
    hered
    1.21
    Activations Density 1.662%

    No Known Activations