© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 2-RES-SM
    11. 13289
    Prev
    Next
    INDEX
    Explanations

    phrases related to legal processes and sentencing durations

    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/2-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.2.hook_resid_post
    Hook Layer
    2
    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
    342
    +0.15
    0.9%
    220
    +0.13
    0.8%
    69
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    220
    +0.15
    0.05
    130
    +0.13
    0.10
    317
    +0.12
    0.08
    Negative Logits
    ¹
    -1.95
    µ
    -1.78
    Ļ
    -1.78
    ¼
    -1.76
    oque
    -1.67
    ²
    -1.67
    ³
    -1.67
    /.
    -1.61
    eanor
    -1.52
    «
    -1.49
    POSITIVE LOGITS
     jail
    1.44
    Expected
    1.43
     thanks
    1.42
    CLUD
    1.38
     ago
    1.36
     apprent
    1.36
    lost
    1.34
     deaths
    1.31
    eligible
    1.30
     women
    1.29
    Activations Density 2.035%

    No Known Activations