© 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. 11623
    Prev
    Next
    INDEX
    Explanations

    phrases indicating certainty or emphasis

    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
    21
    +0.17
    1.0%
    78
    +0.13
    0.7%
    47
    +0.11
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    242
    +0.17
    0.07
    402
    +0.13
    0.09
    257
    +0.11
    0.07
    Negative Logits
    Ļ
    -3.42
    Ĩ
    -3.28
    Ģ
    -2.99
    Ī
    -2.97
    Ħ
    -2.96
    ±
    -2.93
    ĭ
    -2.91
    ¾
    -2.90
    ľ
    -2.89
    ĨĴ
    -2.80
    POSITIVE LOGITS
    ifies
    2.67
     exists
    2.50
     existed
    2.30
     reflects
    2.30
     happens
    2.20
     disappears
    2.20
     incorporates
    2.18
     appears
    2.17
     contains
    2.16
     precludes
    2.07
    Activations Density 0.638%

    No Known Activations