© 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. MLP Post
    9. 2-MLP-SM
    11. 26713
    Prev
    Next
    INDEX
    Explanations

    phrases that indicate evaluation or critique of performance and capabilities

    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
    192
    +0.12
    0.6%
    176
    +0.12
    0.6%
    414
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    376
    +0.12
    0.06
    363
    +0.12
    0.06
    499
    +0.11
    0.05
    Negative Logits
     hypotheses
    -1.70
     reasons
    -1.51
    rox
    -1.50
    ity
    -1.49
    rior
    -1.42
     Biological
    -1.38
    urane
    -1.36
     hypothesis
    -1.35
    INGS
    -1.33
     assumptions
    -1.32
    POSITIVE LOGITS
     Drive
    1.63
    ograf
    1.61
    ↵
    1.59
    ÃŃa
    1.55
    ág
    1.49
    ière
    1.48
    olta
    1.48
    films
    1.48
    esan
    1.47
    ourse
    1.46
    Activations Density 0.745%

    No Known Activations