© 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 EvalsExportsinterp-engineNEWAPI 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. 26857
    Prev
    Next
    INDEX
    Explanations

    terms related to very high levels of intensity or quality

    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
    308
    +0.16
    0.9%
    375
    +0.13
    0.7%
    130
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    375
    +0.16
    0.02
    308
    +0.13
    0.01
    90
    +0.13
    0.02
    Negative Logits
    able
    -1.93
     aloud
    -1.76
     omitted
    -1.58
    :&
    -1.54
    out
    -1.47
     slightest
    -1.40
     forgiven
    -1.39
    oret
    -1.39
     excuses
    -1.39
    gered
    -1.37
    POSITIVE LOGITS
    ·
    1.93
    ·¸
    1.83
    µ
    1.80
    raviolet
    1.77
    »
    1.77
    ī
    1.66
    VIEW
    1.65
    boro
    1.51
    ģ
    1.51
    1.50
    Activations Density 0.164%

    No Known Activations