© 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. Residuals
    9. 4-RES-SM
    11. 12708
    Prev
    Next
    INDEX
    Explanations

    expressions indicating past experiences or behaviors and their evaluations

    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
    294
    +0.13
    0.7%
    117
    +0.12
    0.7%
    161
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    420
    +0.13
    0.07
    161
    +0.12
    0.06
    498
    +0.11
    0.08
    Negative Logits
    ĥ½
    -2.65
    ľ
    -2.58
    £
    -2.50
    ĭ
    -2.50
    Ļª
    -2.36
    ³
    -2.30
    ij
    -2.18
    ĵ
    -2.17
    Ĭ
    -2.15
    Ļ
    -2.13
    POSITIVE LOGITS
     imagined
    1.92
     did
    1.78
     accustomed
    1.77
     normally
    1.74
     expected
    1.69
     thought
    1.67
     hoped
    1.66
     envisioned
    1.62
     supposed
    1.59
     originally
    1.58
    Activations Density 0.721%

    No Known Activations