© 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. 2-RES-SM
    11. 10326
    Prev
    Next
    INDEX
    Explanations

    technical terms and commands relevant to software configuration and development processes

    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
    156
    +0.31
    1.8%
    240
    +0.16
    0.9%
    43
    +0.15
    0.9%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    92
    +0.31
    0.08
    410
    +0.16
    0.10
    43
    +0.15
    0.06
    Negative Logits
    alus
    -1.95
    vere
    -1.77
    yet
    -1.63
    )",
    -1.49
    aeda
    -1.47
    )"
    -1.44
    lasting
    -1.41
    folk
    -1.38
    prototype
    -1.32
    traits
    -1.32
    POSITIVE LOGITS
    ²
    1.98
    °
    1.97
    ¢
    1.95
    Īĺ
    1.94
    ¸
    1.92
    Į
    1.91
    IJ
    1.90
    Ģ
    1.88
    ¶
    1.85
    ħ
    1.82
    Activations Density 0.642%

    No Known Activations