© 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. 1-RES-SM
    11. 27755
    Prev
    Next
    INDEX
    Explanations

    references to specific chemical compounds and their molecular structures

    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/1-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.1.hook_resid_post
    Hook Layer
    1
    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
    111
    +0.15
    0.9%
    13
    +0.13
    0.8%
    410
    +0.13
    0.8%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    13
    +0.15
    0.01
    2
    +0.13
    0.01
    466
    +0.13
    0.01
    Negative Logits
     forth
    -1.73
     evaluating
    -1.65
     participating
    -1.62
     caring
    -1.59
     ready
    -1.59
    ?”
    -1.56
     '</
    -1.54
    fected
    -1.52
     noticing
    -1.51
     your
    -1.49
    POSITIVE LOGITS
    ĻĤ
    3.78
    ¬
    3.74
    ¥
    3.67
    §
    3.65
    £
    3.61
    ¢
    3.54
    ¿½
    3.54
    Ģ
    3.53
    ľ
    3.52
    3.47
    Activations Density 0.047%

    No Known Activations