Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    3. Pythia-70M-Deduped
    4. Residuals
    5. 2-RES-SM
    6. 11832
    Prev
    Next
    INDEX
    Explanations

    technical terms related to geometry and spatial calculations

    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
    +0.21
    1.2%
    17
    +0.17
    1.0%
    60
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    17
    +0.21
    0.03
    47
    +0.17
    0.10
    252
    +0.12
    0.07
    Negative Logits
     affiliate
    -1.85
     administr
    -1.73
     abst
    -1.70
     administrative
    -1.69
     opportun
    -1.68
     antiviral
    -1.63
     privileges
    -1.63
     oath
    -1.62
    ÂĢÂĤ
    -1.62
     provision
    -1.60
    POSITIVE LOGITS
     perpendicular
    2.26
     pixels
    2.03
    ygon
    1.98
     radius
    1.93
     Pixel
    1.93
    ixel
    1.91
     pixel
    1.87
     texture
    1.87
    brush
    1.83
    uler
    1.82
    Activations Density 1.233%

    No Known Activations