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. 11698
    Prev
    Next
    INDEX
    Explanations

    references to procedural or regulatory steps in contexts such as legal or construction processes

    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
    +0.74
    6.0%
    23
    +0.21
    1.7%
    478
    +0.11
    0.9%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    23
    +0.74
    0.20
    478
    +0.21
    -0.02
    240
    +0.11
    -0.05
    Negative Logits
    аÑħ
    -1.58
    :`
    -1.50
    à½
    -1.46
    ி
    -1.45
    à¯
    -1.44
    à®ķ
    -1.43
    ressor
    -1.40
    Õ¥
    -1.39
    à²
    -1.38
    á̏
    -1.38
    POSITIVE LOGITS
     respect
    1.43
     blunt
    1.23
     how
    1.22
     basic
    1.20
     another
    1.20
     amounts
    1.18
     gaug
    1.18
    ching
    1.17
     precise
    1.16
     suppose
    1.16
    Activations Density 4.359%

    No Known Activations