© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 10887
    Prev
    Next
    INDEX
    Explanations

    phrases indicating desires, intentions, and requests for knowledge or action

    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
    +0.23
    1.3%
    17
    +0.18
    1.0%
    323
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    263
    +0.23
    0.11
    323
    +0.18
    0.04
    239
    +0.13
    0.07
    Negative Logits
    ĵ
    -3.31
    Ļª
    -3.10
    ¸
    -3.08
    į
    -3.02
    ĥ
    -2.97
    §
    -2.95
    ħ
    -2.92
    ģ
    -2.83
    Ŀ
    -2.81
    ŀ
    -2.80
    POSITIVE LOGITS
     photographed
    1.61
     sure
    1.53
     hosted
    1.50
     distributed
    1.41
     won
    1.40
     caffe
    1.39
     streaming
    1.37
     things
    1.35
     /
    1.35
     logged
    1.28
    Activations Density 0.737%

    No Known Activations