© 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. 4-RES-SM
    11. 10824
    Prev
    Next
    INDEX
    Explanations

    instances of the word "follow."

    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
    306
    +0.14
    0.8%
    380
    +0.13
    0.7%
    263
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    380
    +0.14
    0.04
    485
    +0.13
    0.02
    429
    +0.12
    0.02
    Negative Logits
    ī
    -2.52
    İ
    -2.46
    ģ
    -2.37
    ļ
    -2.27
    ħ
    -2.09
    º
    -2.09
    Ľ
    -2.08
    «
    -2.08
    ¥
    -2.01
    ¶
    -2.01
    POSITIVE LOGITS
     suit
    2.87
     footsteps
    2.30
    emente
    2.22
     closely
    2.18
     instructions
    2.02
     orders
    1.90
     up
    1.74
     conventions
    1.74
     directions
    1.58
    noon
    1.55
    Activations Density 0.272%

    No Known Activations