© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    2. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    3. Pythia-70M-Deduped
    4. Residuals
    5. 2-RES-SM
    6. 29907
    Prev
    Next
    INDEX
    Explanations

    verbs related to reliability and effectiveness in actions or 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
    263
    +0.17
    0.9%
    455
    +0.15
    0.9%
    500
    +0.15
    0.9%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    500
    +0.17
    0.09
    476
    +0.15
    0.09
    263
    +0.15
    0.09
    Negative Logits
    Ļª
    -1.74
    ļ
    -1.72
    ¦
    -1.72
    ij
    -1.66
    Ĺ
    -1.59
    ி
    -1.58
    ĺ
    -1.57
    ħ
    -1.51
    §
    -1.51
    ´
    -1.48
    POSITIVE LOGITS
     recreate
    1.93
     allocate
    1.86
    fiddle
    1.81
     reproduce
    1.75
    IFY
    1.74
     manage
    1.73
    ocate
    1.68
    itate
    1.62
    develop
    1.58
     simulate
    1.58
    Activations Density 1.124%

    No Known Activations