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

    Neuronpedia

    Jacobian LensNEW
    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. 4-RES-SM
    11. 12893
    Prev
    Next
    INDEX
    Explanations

    phrases or words indicating various forms of capability or ability

    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/4-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.4.hook_resid_post
    Hook Layer
    4
    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
    348
    +0.17
    1.0%
    78
    +0.15
    0.9%
    9
    +0.11
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    352
    +0.17
    0.02
    122
    +0.15
    0.03
    66
    +0.11
    0.03
    Negative Logits
    ché
    -1.39
    friends
    -1.37
     thought
    -1.33
    etheless
    -1.32
     plans
    -1.28
    iffe
    -1.27
     weekend
    -1.26
     sitting
    -1.25
    friend
    -1.24
     caso
    -1.23
    POSITIVE LOGITS
     to
    1.70
    ility
    1.66
    urable
    1.65
     conferred
    1.59
    ometer
    1.57
    stone
    1.53
    arynge
    1.47
    burg
    1.45
    finder
    1.45
    ence
    1.44
    Activations Density 0.153%

    No Known Activations