© 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
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 2-RES-SM
    11. 28687
    Prev
    Next
    INDEX
    Explanations

    code-related terminology and structures in programming languages

    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
    47
    +0.16
    0.9%
    63
    +0.14
    0.8%
    479
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    444
    +0.16
    0.00
    63
    +0.14
    0.06
    47
    +0.13
    0.14
    Negative Logits
    ¿½
    -2.05
    ľĵ
    -1.93
    ®
    -1.78
    ¨
    -1.78
    ¿
    -1.75
    ª
    -1.73
    º
    -1.72
    ©
    -1.67
    ²
    -1.66
    Į
    -1.65
    POSITIVE LOGITS
     such
    1.56
     simplest
    1.46
     nerves
    1.43
    orate
    1.38
    ermine
    1.37
    arer
    1.35
     _
    1.34
     underlying
    1.30
    ples
    1.28
    ody
    1.27
    Activations Density 4.308%

    No Known Activations