© 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. 11782
    Prev
    Next
    INDEX
    Explanations

    sections of code related to structured data management and software framework references

    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
    78
    +0.28
    1.7%
    126
    +0.15
    0.9%
    191
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    191
    +0.28
    0.03
    126
    +0.15
    0.01
    290
    +0.10
    0.02
    Negative Logits
     disadvantage
    -1.51
     misdem
    -1.36
    ynaptic
    -1.36
    mers
    -1.35
     liability
    -1.30
     drought
    -1.27
     remind
    -1.26
    eligible
    -1.26
     reminder
    -1.25
    umab
    -1.24
    POSITIVE LOGITS
    ¤
    2.78
    ĭ
    2.71
    ³
    2.65
    ŀ
    2.64
    §
    2.63
    İ
    2.61
      ↵
    2.53
    ·
    2.52
    <|outofrange|>
    2.52
    ↵↵↵
    2.52
    Activations Density 0.045%

    No Known Activations