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

    Neuronpedia

    Jacobian LensNEW
    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. 24826
    Prev
    Next
    INDEX
    Explanations

    discussions about access and opportunities within communities

    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
    330
    +0.13
    0.8%
    57
    +0.13
    0.7%
    179
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    330
    +0.13
    0.13
    223
    +0.13
    0.06
    104
    +0.12
    0.08
    Negative Logits
     convenient
    -1.47
     whisk
    -1.47
     reach
    -1.36
    abol
    -1.36
     advantageous
    -1.35
     territory
    -1.35
     attract
    -1.35
     loyalty
    -1.34
     favourable
    -1.34
     fairness
    -1.32
    POSITIVE LOGITS
    ĻĤ
    3.49
    ©
    3.21
    ±
    2.97
    ·¸
    2.79
    ģ
    2.77
    °
    2.58
    IJ
    2.56
    Ģ
    2.48
    ĭ
    2.43
    Ĵ
    2.41
    Activations Density 2.189%

    No Known Activations