© 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. 11622
    Prev
    Next
    INDEX
    Explanations

    quantifiable assessments or expressions of significance

    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
    172
    +0.14
    0.8%
    49
    +0.12
    0.7%
    366
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    137
    +0.14
    0.06
    56
    +0.12
    0.02
    172
    +0.12
    0.04
    Negative Logits
    ij
    -2.70
    ·¸
    -2.69
    ¸
    -2.56
    ¦
    -2.54
    »¿
    -2.49
    İ
    -2.48
    ·
    -2.47
    «
    -2.43
    ³
    -2.41
    ħ
    -2.39
    POSITIVE LOGITS
     difference
    2.37
     comeback
    2.07
     adjustments
    2.00
     happen
    1.98
     decisions
    1.79
     mistake
    1.77
     contribution
    1.77
    difference
    1.76
    case
    1.74
     statement
    1.72
    Activations Density 0.504%

    No Known Activations