© 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. Residuals
    9. 2-RES-SM
    11. 30175
    Prev
    Next
    INDEX
    Explanations

    processes related to analysis and evaluation

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    ctigges/pythia-70m-deduped__res-sm_processed/2-res-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_resid_post
    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
    263
    +0.16
    0.9%
    95
    +0.15
    0.9%
    16
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    16
    +0.16
    0.12
    412
    +0.15
    0.15
    148
    +0.13
    0.04
    Negative Logits
    ŀ
    -3.42
    ģ
    -3.42
    Ŀ
    -3.40
    ¯
    -3.35
    ĥ
    -3.35
    ·¸
    -3.22
    ĻĤ
    -3.21
    ¾
    -3.18
    §
    -3.16
    «
    -3.16
    POSITIVE LOGITS
    obox
    1.80
    ently
    1.72
    mouth
    1.72
     validity
    1.59
     lately
    1.46
    tract
    1.46
    iquity
    1.45
    liqu
    1.44
    (),
    1.44
     spelling
    1.44
    Activations Density 2.302%

    No Known Activations