© 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. 10445
    Prev
    Next
    INDEX
    Explanations

    patterns or sequences of numeric values combined with arithmetic operations

    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
    125
    +0.15
    0.9%
    135
    +0.11
    0.6%
    168
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    125
    +0.15
    0.02
    26
    +0.11
    0.01
    168
    +0.11
    0.02
    Negative Logits
     proportions
    -1.59
    piece
    -1.52
    lessness
    -1.50
     involves
    -1.50
     mine
    -1.47
    pieces
    -1.45
     attention
    -1.44
    .[]{
    -1.43
    ose
    -1.41
    ings
    -1.41
    POSITIVE LOGITS
    ¡
    3.67
    ĨĴ
    3.58
    Īĺ
    3.56
    ĥ½
    3.54
    <|outofrange|>
    3.44
    3.44
                ↵
    3.44
    3.44
    ↵
    3.44
    ↵
    3.44
    Activations Density 0.020%

    No Known Activations