© 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. 10785
    Prev
    Next
    INDEX
    Explanations

    the word "most" in various contexts

    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
    8
    +0.12
    0.7%
    102
    +0.11
    0.7%
    284
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    102
    +0.12
    0.04
    284
    +0.11
    0.03
    8
    +0.11
    0.03
    Negative Logits
    set
    -1.47
    UMENT
    -1.47
    hetti
    -1.45
    abad
    -1.44
    chell
    -1.43
    ulator
    -1.42
    oyle
    -1.42
    ettle
    -1.41
    wards
    -1.41
     modo
    -1.39
    POSITIVE LOGITS
     importantly
    3.42
     notably
    2.78
     likely
    2.74
     important
    2.33
     commonly
    2.16
     recently
    2.14
     prominently
    2.13
     of
    2.10
     famously
    2.08
     notable
    1.97
    Activations Density 0.103%

    No Known Activations