© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. MLP Post
    9. 2-MLP-SM
    11. 22377
    Prev
    Next
    INDEX
    Explanations

    quotation marks or dialog in the text

    oai_token-act-pair · gpt-4o-mini
    Triggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    ctigges/pythia-70m-deduped__mlp-sm_processed/2-mlp-sm
    Prompts (Dashboard)
    32,768 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.2.hook_mlp_out
    Hook Layer
    2
    Architecture
    standard
    Context Size
    128
    Dataset
    EleutherAI/the_pile_deduplicated
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Neuron Alignment
    Index
    Value
    % of L₁
    76
    +0.12
    0.6%
    161
    +0.11
    0.6%
    292
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    314
    +0.12
    0.01
    300
    +0.11
    0.01
    237
    +0.11
    0.01
    Negative Logits
     summary
    -1.70
    ardo
    -1.60
    inn
    -1.59
    rectomy
    -1.55
    áĢº
    -1.54
    liness
    -1.53
    acht
    -1.52
     breakdown
    -1.50
     chance
    -1.49
    nex
    -1.48
    POSITIVE LOGITS
    ĥ½
    1.93
    ulates
    1.75
    ĻĤ
    1.60
    othes
    1.44
    ¼
    1.43
    ETHOD
    1.42
    ifferent
    1.41
    WITH
    1.41
    İ
    1.41
    izumab
    1.37
    Activations Density 0.004%

    No Known Activations