© 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. 28208
    Prev
    Next
    INDEX
    Explanations

    instances of the word "har" which signifies possession or ownership

    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
    376
    +0.16
    0.9%
    111
    +0.14
    0.8%
    356
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    111
    +0.16
    0.01
    138
    +0.14
    0.01
    450
    +0.12
    0.01
    Negative Logits
    Ĥ
    -1.73
    ĥ
    -1.56
    ĩ
    -1.52
    ethylene
    -1.52
    ANC
    -1.52
     welcome
    -1.49
     presence
    -1.40
    GFR
    -1.40
    Jet
    -1.37
     slightest
    -1.37
    POSITIVE LOGITS
    ppo
    1.79
     than
    1.71
    igan
    1.70
    leys
    1.65
    igraphy
    1.65
    ousel
    1.64
     annealing
    1.64
    ainties
    1.62
    shire
    1.59
    pending
    1.59
    Activations Density 0.010%

    No Known Activations