© 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. 2-RES-SM
    11. 10512
    Prev
    Next
    INDEX
    Explanations

    numerical dimensions and specifications

    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
    77
    +0.18
    1.0%
    409
    +0.17
    1.0%
    115
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    307
    +0.18
    0.05
    477
    +0.17
    0.06
    82
    +0.13
    0.04
    Negative Logits
    °
    -1.68
    Ĥ
    -1.58
    ĻĤ
    -1.57
    unreadable
    -1.54
     suppose
    -1.49
    Īĺ
    -1.48
    ¹
    -1.46
     nation
    -1.39
    ±
    -1.38
     contention
    -1.38
    POSITIVE LOGITS
     Minutes
    1.99
    Hz
    1.89
    shire
    1.79
     minutes
    1.73
    oon
    1.70
     Springs
    1.68
     onward
    1.61
     hours
    1.56
    som
    1.55
     Hz
    1.54
    Activations Density 0.218%

    No Known Activations