© 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. 11085
    Prev
    Next
    INDEX
    Explanations

    code-related structures and elements

    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
    302
    +0.18
    1.0%
    76
    +0.16
    0.9%
    419
    +0.15
    0.9%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    302
    +0.18
    0.09
    377
    +0.16
    0.10
    140
    +0.15
    0.08
    Negative Logits
    ever
    -1.52
     Bengal
    -1.42
    ickets
    -1.34
     Sierra
    -1.32
     schools
    -1.32
     retirement
    -1.32
    died
    -1.30
     owners
    -1.28
     Indonesia
    -1.28
     safety
    -1.28
    POSITIVE LOGITS
    >:
    1.57
    blockList
    1.54
    ostream
    1.52
    $/
    1.46
    ools
    1.44
     xmlns
    1.43
    ário
    1.42
    ipeline
    1.38
    imon
    1.37
    $^
    1.37
    Activations Density 5.776%

    No Known Activations