© 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. 4-RES-SM
    11. 11993
    Prev
    Next
    INDEX
    Explanations

    the word "known" along with variations of its usage and contextual associations

    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
    +0.15
    0.9%
    183
    +0.12
    0.7%
    197
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    420
    +0.15
    0.02
    304
    +0.12
    0.02
    330
    +0.12
    0.02
    Negative Logits
    «
    -4.98
    º
    -4.65
    ¥
    -4.63
    µ
    -4.56
    Īĺ
    -4.50
    Ń
    -4.49
    ı
    -4.48
    ī
    -4.36
    Ļ
    -4.33
    ĺ
    -4.26
    POSITIVE LOGITS
     thereto
    2.22
     to
    2.02
     therefrom
    1.84
     as
    1.74
     collectively
    1.65
     from
    1.63
     compositions
    1.59
     [@
    1.56
     beforehand
    1.55
     internationally
    1.52
    Activations Density 0.085%

    No Known Activations