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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. DeepSeek-R1-Dist-Llama-8B
    3. 15-LLAMASCOPE-SLIMPJ-RES-32K
    4. 1642
    Prev
    Next
    INDEX
    Explanations

    negative phrases or expressions indicating necessity or obligation

    oai_token-act-pair · gpt-4o-mini
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    fnlp/Llama-Scope-R1-Distill/800M-Slimpajama-0-OpenR1-Math-220k/L15R
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Hzfinfdu/SlimPajama-3B
    Features
    32,768
    Data Type
    float32
    Hook Name
    blocks.15.hook_resid_post
    Hook Layer
    15
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    cerebras/SlimPajama-627B
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    assis
    -0.14
    аÑĤем
    -0.14
    åħį
    -0.14
     indu
    -0.14
    KO
    -0.14
    stag
    -0.14
    ingular
    -0.14
    uro
    -0.14
    ruba
    -0.13
    _frm
    -0.13
    POSITIVE LOGITS
     worry
    0.25
     worried
    0.17
    rypton
    0.16
     worrying
    0.16
    abal
    0.16
     concern
    0.15
     worries
    0.15
     fret
    0.15
     necessarily
    0.14
    æĭħ
    0.14
    Activations Density 0.047%

    No Known Activations