© 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. 30624
    Prev
    Next
    INDEX
    Explanations

    punctuation marks, particularly periods

    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
    203
    +0.18
    1.0%
    258
    +0.17
    0.9%
    369
    +0.14
    0.8%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    60
    +0.18
    0.04
    203
    +0.17
    0.06
    432
    +0.14
    0.04
    Negative Logits
    athi
    -1.86
    oelectric
    -1.76
    ceedings
    -1.69
    ÅĽci
    -1.69
    "/>
    -1.59
    fiction
    -1.58
    ert
    -1.58
    ţi
    -1.55
    olta
    -1.54
    uchi
    -1.51
    POSITIVE LOGITS
    Ĥ
    2.58
    §
    2.33
    ģ
    2.22
    ©
    2.14
    ¾
    2.00
    Į
    1.97
    ĥ½
    1.96
    ¶
    1.91
    ľĵ
    1.87
    ħ
    1.86
    Activations Density 0.442%

    No Known Activations