© 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. 31703
    Prev
    Next
    INDEX
    Explanations

    various types of names and variables in code-like formats

    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
    427
    +0.17
    1.0%
    231
    +0.15
    0.9%
    125
    +0.15
    0.9%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    427
    +0.17
    0.09
    125
    +0.15
    0.07
    425
    +0.15
    0.13
    Negative Logits
    ensit
    -1.75
    license
    -1.56
    ickets
    -1.37
    LINK
    -1.36
    ![
    -1.33
    !)
    -1.30
    *.*
    -1.28
     minors
    -1.26
    orage
    -1.26
    wrap
    -1.26
    POSITIVE LOGITS
    ço
    1.67
    stitutional
    1.43
     whom
    1.40
    xico
    1.40
    esque
    1.39
    iast
    1.39
     indictment
    1.36
    iÄĩ
    1.34
    cephal
    1.33
    Äĩ
    1.31
    Activations Density 3.670%

    No Known Activations