© 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. 4-RES-SM
    11. 14441
    Prev
    Next
    INDEX
    Explanations

    scientific citations and references

    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/4-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.4.hook_resid_post
    Hook Layer
    4
    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
    21
    +0.25
    1.4%
    255
    +0.15
    0.9%
    329
    +0.15
    0.8%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    255
    +0.25
    0.00
    21
    +0.15
    0.02
    333
    +0.15
    0.03
    Negative Logits
    terday
    -1.77
    atically
    -1.67
    tered
    -1.61
    Äĩ
    -1.60
    mere
    -1.54
     fare
    -1.53
     Lear
    -1.52
     sanity
    -1.48
    inally
    -1.47
    ized
    -1.46
    POSITIVE LOGITS
    Ģ
    3.64
    £
    3.37
    ¡
    3.36
    ±
    3.32
    ĻĤ
    3.32
    ³
    3.31
    ©
    3.17
    ij
    3.14
    Ī
    3.13
    ĥ
    3.13
    Activations Density 0.200%

    No Known Activations