© 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. 2-RES-SM
    11. 26957
    Prev
    Next
    INDEX
    Explanations

    terms related to membrane structures and their properties

    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
    156
    +0.23
    1.4%
    476
    +0.13
    0.8%
    304
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    306
    +0.23
    0.01
    304
    +0.13
    0.01
    173
    +0.11
    0.01
    Negative Logits
    olic
    -1.85
     pride
    -1.80
    cember
    -1.72
    hhhh
    -1.67
    iamo
    -1.63
     birthday
    -1.62
    arcer
    -1.58
    enton
    -1.56
    cester
    -1.55
     girlfriend
    -1.55
    POSITIVE LOGITS
    works
    1.98
    atics
    1.80
     composed
    1.75
    imeters
    1.69
    bridge
    1.64
    plate
    1.61
    stalk
    1.59
     piercing
    1.58
    walker
    1.52
    orphic
    1.50
    Activations Density 0.036%

    No Known Activations