© 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
    2. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    3. Pythia-70M-Deduped
    4. MLP Post
    5. 0-MLP-SM
    6. 26787
    Prev
    Next
    INDEX
    Explanations

    instances of repetitive conjunctions, specifically "and"

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    ctigges/pythia-70m-deduped__mlp-sm_processed/0-mlp-sm
    Prompts (Dashboard)
    32,768 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.0.hook_mlp_out
    Hook Layer
    0
    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
    +0.14
    0.8%
    194
    +0.12
    0.7%
    93
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    494
    +0.14
    0.45
    209
    +0.12
    0.29
    402
    +0.11
    0.26
    Negative Logits
    istical
    -1.56
    semble
    -1.53
    holds
    -1.49
    sembl
    -1.48
    lements
    -1.48
    build
    -1.45
    heimer
    -1.44
     joint
    -1.44
    leg
    -1.42
    go
    -1.42
    POSITIVE LOGITS
    ĩ
    1.50
    ”)
    1.39
    hydr
    1.35
    ATP
    1.31
    ingly
    1.31
    rowning
    1.29
     interrupted
    1.29
    cient
    1.29
     awake
    1.29
     brakes
    1.29
    Activations Density 1.588%

    No Known Activations