© 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 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. 10708
    Prev
    Next
    INDEX
    Explanations

    negative sentiment or expressions of discontent

    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
    21
    +0.23
    1.5%
    78
    +0.22
    1.4%
    156
    +0.22
    1.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    203
    +0.23
    0.05
    448
    +0.22
    0.07
    450
    +0.22
    0.03
    Negative Logits
     interest
    -1.70
    LLY
    -1.40
    alities
    -1.39
    ctions
    -1.39
     scores
    -1.38
    autres
    -1.37
     outcome
    -1.36
     consent
    -1.34
     fare
    -1.33
     Settings
    -1.33
    POSITIVE LOGITS
    Ķ
    2.46
    Ĥ
    2.33
    °
    2.29
    ©
    2.29
    ·
    2.25
    ī
    2.23
    Ĥ¬
    2.19
    ¢
    2.17
    IJ
    2.15
    »
    2.14
    Activations Density 0.211%

    No Known Activations