© 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. 4-RES-SM
    11. 11654
    Prev
    Next
    INDEX
    Explanations

    phrases focusing on ease of use and simplifying complex processes

    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/4-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.4.hook_resid_post
    Hook Layer
    4
    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
    387
    +0.14
    0.8%
    2
    +0.12
    0.7%
    185
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    70
    +0.14
    0.07
    387
    +0.12
    0.07
    144
    +0.12
    0.04
    Negative Logits
    ĻĤ
    -5.54
    Ħ
    -5.28
    İ
    -5.05
    ¿½
    -4.99
    ±
    -4.89
    ¯
    -4.87
    £
    -4.78
    į
    -4.75
    »¿
    -4.73
    ¸
    -4.72
    POSITIVE LOGITS
     easier
    2.94
     safer
    2.60
     harder
    2.45
     attractive
    2.39
     easy
    2.36
     obsolete
    2.30
     impossible
    2.24
     difficult
    2.24
     possible
    2.24
     extremely
    2.23
    Activations Density 0.664%

    No Known Activations