© 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. 26638
    Prev
    Next
    INDEX
    Explanations

    forms of the verbs "can," "is," and "will."

    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
    156
    +0.20
    1.1%
    23
    +0.15
    0.8%
    12
    +0.14
    0.8%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    12
    +0.20
    0.07
    501
    +0.15
    0.05
    367
    +0.14
    0.02
    Negative Logits
    č↵
    -3.65
    č↵
    -3.65
    ↵  âĢĥ
    -3.65
    ↵
    -3.65
    ↵↵
    -3.65
    -3.65
    ↵
    -3.65
    ↵
    -3.65
    -3.65
    -3.65
    POSITIVE LOGITS
     Rodriguez
    1.54
    oms
    1.41
     tasted
    1.38
     Nil
    1.35
    ont
    1.35
    anian
    1.34
    !\
    1.34
    !)
    1.33
     Ramirez
    1.33
    BIT
    1.32
    Activations Density 0.741%

    No Known Activations