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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 2-RES-SM
    11. 30181
    Prev
    Next
    INDEX
    Explanations

    phrases indicating quantitative relationships or proportions

    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
    244
    +0.13
    0.7%
    480
    +0.12
    0.7%
    302
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    244
    +0.13
    0.06
    480
    +0.12
    0.05
    152
    +0.11
    0.04
    Negative Logits
     heartbeat
    -1.43
    idone
    -1.36
    ubicin
    -1.35
    rah
    -1.35
    disambiguation
    -1.35
    quart
    -1.32
    bye
    -1.28
    ACE
    -1.27
    ureus
    -1.27
    asan
    -1.24
    POSITIVE LOGITS
     brackets
    1.57
    itals
    1.54
     resources
    1.46
     these
    1.46
    tons
    1.40
     APPEALS
    1.39
     theirs
    1.36
     particular
    1.33
    mund
    1.32
     ?>
    1.32
    Activations Density 0.158%

    No Known Activations