© 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. 12364
    Prev
    Next
    INDEX
    Explanations

    occurrences of the letter 'o' and variations in its form

    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
    156
    +0.18
    1.0%
    130
    +0.16
    0.9%
    372
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    130
    +0.18
    0.06
    396
    +0.16
    0.03
    410
    +0.13
    0.03
    Negative Logits
    mable
    -2.01
    day
    -1.93
    night
    -1.79
    nis
    -1.75
    man
    -1.72
    days
    -1.70
    woman
    -1.68
    bread
    -1.68
     msgid
    -1.65
    nar
    -1.64
    POSITIVE LOGITS
    ftware
    2.77
    ÅĽÄĩ
    2.72
    cean
    2.50
    ÅĽci
    2.47
    ji
    2.46
    olean
    2.35
    vel
    2.26
    ops
    2.06
    zzi
    2.05
    eba
    1.99
    Activations Density 0.365%

    No Known Activations