© 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. 2-RES-SM
    11. 10799
    Prev
    Next
    INDEX
    Explanations

    descriptive adjectives pertaining to style and color

    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
    169
    +0.13
    0.8%
    56
    +0.11
    0.6%
    231
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    148
    +0.13
    0.03
    198
    +0.11
    0.08
    169
    +0.11
    0.09
    Negative Logits
    izumab
    -1.49
     impaired
    -1.44
    Ïĩ
    -1.44
    ach
    -1.44
    minent
    -1.43
     fatal
    -1.42
    ori
    -1.40
     killed
    -1.38
     adversely
    -1.38
     exacerb
    -1.37
    POSITIVE LOGITS
    ĻĤ
    2.03
    »¿
    1.98
    ¿½
    1.85
    cloth
    1.84
     underneath
    1.79
    bows
    1.77
     ones
    1.76
     arches
    1.75
     outdoors
    1.73
     leather
    1.71
    Activations Density 1.486%

    No Known Activations