© 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. 4-RES-SM
    11. 12312
    Prev
    Next
    INDEX
    Explanations

    numerical values that appear in the context of statistical or quantitative data

    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/4-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.4.hook_resid_post
    Hook Layer
    4
    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
    255
    +0.16
    0.9%
    21
    +0.14
    0.9%
    78
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    200
    +0.16
    0.04
    133
    +0.14
    0.04
    255
    +0.10
    0.00
    Negative Logits
     chapter
    -1.53
     department
    -1.44
     caption
    -1.38
     closer
    -1.35
     scene
    -1.30
     pause
    -1.30
     proper
    -1.29
     volume
    -1.29
     classes
    -1.28
     serial
    -1.27
    POSITIVE LOGITS
    }_{-
    1.73
    5
    1.64
    6
    1.59
    7
    1.55
    8
    1.52
    9
    1.51
    0
    1.48
    aiver
    1.47
    67
    1.46
    idegger
    1.45
    Activations Density 0.162%

    No Known Activations