© 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. 26950
    Prev
    Next
    INDEX
    Explanations

    references to file paths and directory structures in code

    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
    +0.16
    0.9%
    110
    +0.13
    0.8%
    502
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    110
    +0.16
    0.02
    17
    +0.13
    0.01
    290
    +0.13
    0.03
    Negative Logits
    ľĵ
    -2.53
    ®
    -2.52
    «
    -2.33
    ½
    -2.31
    ķ
    -2.27
     regard
    -2.27
    ŀ
    -2.19
    ĥ½
    -2.15
    ļ
    -2.13
    Ń
    -2.09
    POSITIVE LOGITS
    finder
    2.08
    walker
    1.84
    ways
    1.76
    way
    1.68
     conserved
    1.56
    olog
    1.53
    vation
    1.53
    ="/
    1.53
    ophys
    1.51
    walk
    1.50
    Activations Density 0.128%

    No Known Activations