© 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. 12699
    Prev
    Next
    INDEX
    Explanations

    technical details or specifications related to software and programming

    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
    263
    +0.17
    1.0%
    161
    +0.14
    0.8%
    17
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    20
    +0.17
    0.15
    451
    +0.14
    0.05
    223
    +0.12
    0.06
    Negative Logits
    empre
    -1.58
    )):
    -1.56
    ently
    -1.55
    agma
    -1.53
    aterally
    -1.49
    /#
    -1.48
    )),
    -1.47
     purposes
    -1.44
     detriment
    -1.41
    ovan
    -1.40
    POSITIVE LOGITS
    kerchief
    1.46
     friends
    1.33
     administration
    1.32
     led
    1.28
     Administration
    1.28
    atar
    1.27
     Government
    1.23
     soldiers
    1.22
     party
    1.19
     strangers
    1.19
    Activations Density 4.790%

    No Known Activations