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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & 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. 28559
    Prev
    Next
    INDEX
    Explanations

    information related to government announcements about examinations and public services

    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
    254
    +0.13
    0.7%
    151
    +0.13
    0.7%
    340
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    410
    +0.13
    0.08
    279
    +0.13
    0.07
    271
    +0.12
    -0.00
    Negative Logits
    ÂĹ
    -1.78
     <-
    -1.78
     offseason
    -1.76
    Plaintiffs
    -1.65
     Redistributions
    -1.60
    opically
    -1.60
     NFL
    -1.60
    ller
    -1.59
     preempt
    -1.57
     MLS
    -1.52
    POSITIVE LOGITS
     Tamil
    2.47
     Shah
    2.31
     Kumar
    2.27
     Sri
    2.22
    ’.
    2.15
     Bh
    2.13
     Singh
    2.13
     Raj
    2.10
     Modi
    2.01
     Gandhi
    2.00
    Activations Density 0.936%

    No Known Activations