© 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. 10261
    Prev
    Next
    INDEX
    Explanations

    terms related to movement and motion dynamics

    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
    156
    +0.11
    0.6%
    405
    +0.10
    0.6%
    426
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    287
    +0.11
    0.03
    426
    +0.10
    0.02
    381
    +0.10
    0.02
    Negative Logits
    ľĵ
    -3.14
    ĥ½
    -2.85
    «
    -2.57
    §
    -2.48
    ¸
    -2.41
    ĥ
    -2.30
    ·¸
    -2.29
    »¿
    -2.26
    ĸ´
    -2.23
    ¿½
    -2.19
    POSITIVE LOGITS
     between
    1.93
     toward
    1.92
    mapsto
    1.91
     towards
    1.85
    behind
    1.74
    Handler
    1.71
    gap
    1.68
    ahead
    1.68
     emitted
    1.66
    stream
    1.65
    Activations Density 0.117%

    No Known Activations