© 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. 28233
    Prev
    Next
    INDEX
    Explanations

    elements of HTML structure and formatted data

    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
    320
    +0.20
    1.2%
    47
    +0.15
    0.9%
    312
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    320
    +0.20
    0.13
    360
    +0.15
    0.06
    121
    +0.13
    0.07
    Negative Logits
    ahoma
    -1.60
    oxford
    -1.58
     notices
    -1.54
    eve
    -1.51
     families
    -1.51
    nob
    -1.45
    ively
    -1.44
    >/
    -1.42
     seriously
    -1.41
     annually
    -1.41
    POSITIVE LOGITS
     classifier
    1.62
    °
    1.57
    į
    1.48
    ¥
    1.47
    Ĵ
    1.47
    âĢ¯
    1.43
    ĺ
    1.41
    Ħ
    1.39
    Ī
    1.39
     midfield
    1.38
    Activations Density 5.700%

    No Known Activations