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. 11617
    Prev
    Next
    INDEX
    Explanations

    syntactical structures and definitions in programming or logical constructs

    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
    302
    +0.18
    1.0%
    31
    +0.14
    0.8%
    271
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    7
    +0.18
    0.03
    167
    +0.14
    0.07
    421
    +0.13
    0.04
    Negative Logits
    pora
    -1.60
    uler
    -1.51
     reductase
    -1.48
    ')->
    -1.44
    tte
    -1.40
     %{
    -1.39
     normalization
    -1.37
     transplantation
    -1.36
     productive
    -1.34
     immunodef
    -1.34
    POSITIVE LOGITS
    ¼
    1.88
    ¿½
    1.82
    uously
    1.67
    ¸
    1.49
    cing
    1.45
    ³
    1.44
    apine
    1.42
    akin
    1.40
    ²
    1.37
     lives
    1.33
    Activations Density 3.896%

    No Known Activations