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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. DeepSeek-R1-Dist-Llama-8B
    5. 15-LLAMASCOPE-OPENR1-RES-32K
    7. 1367
    Prev
    Next
    INDEX
    Explanations

    negative qualifiers in reasoning or logic-related contexts

    oai_token-act-pair · gpt-4o-mini
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    fnlp/Llama-Scope-R1-Distill/0-Slimpajama-800M-OpenR1-Math-220k/L15R
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    open-r1/OpenR1-Math-220k
    Features
    32,768
    Data Type
    float32
    Hook Name
    blocks.15.hook_resid_post
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    cerebras/SlimPajama-627B
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    essen
    -0.07
     analog
    -0.06
    ette
    -0.06
    leton
    -0.06
    ä¸įè¦ģ
    -0.06
    Mob
    -0.06
    PURE
    -0.06
     Vác
    -0.06
    le
    -0.06
    uro
    -0.06
    POSITIVE LOGITS
     ones
    0.09
     necessarily
    0.09
     merely
    0.08
    just
    0.06
    itoris
    0.06
    ivor
    0.06
     reserve
    0.06
    udas
    0.06
     just
    0.06
    uali
    0.06
    Activations Density 0.037%

    No Known Activations