© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. MLP Post
    9. 2-MLP-SM
    11. 25878
    Prev
    Next
    INDEX
    Explanations

    the presence of mathematical terms and concepts

    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/2-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.2.hook_mlp_out
    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
    23
    +0.16
    0.9%
    329
    +0.13
    0.7%
    507
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    177
    +0.16
    0.03
    329
    +0.13
    0.02
    411
    +0.13
    0.04
    Negative Logits
    ®
    -2.26
    ¾
    -2.11
    ļ
    -2.06
    ħ
    -2.01
    ½
    -1.99
    ĵ
    -1.98
    µ
    -1.92
    ł
    -1.92
    à°¿
    -1.90
    à±ģ
    -1.87
    POSITIVE LOGITS
     afterward
    1.65
     soon
    1.62
     prayed
    1.62
     promptly
    1.60
     tear
    1.59
     immediately
    1.55
     afterwards
    1.55
     cried
    1.54
     wore
    1.53
    achus
    1.52
    Activations Density 1.796%

    No Known Activations