© 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. OpenMOSS · Llama Scope: SAEs for Llama-3.1-8B
    5. Llama3.1-8B (Base)
    7. Residual Stream
    9. 20-LLAMASCOPE-RES-32K
    11. 30337
    Prev
    Next
    INDEX
    Explanations

    discussions around cooperation, fighting challenges, and effective collective actions against threats

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With LLAMA3.1-8B @ 20-llamascope-res-32k
    Configuration
    fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L20R-8x
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    cerebras/SlimPajama-627B
    Features
    32,768
    Data Type
    bfloat16
    Hook Name
    blocks.20.hook_resid_post
    Hook Layer
    20
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    cerebras/SlimPajama-627B
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     fitte
    -0.20
    enna
    -0.15
    ;amp
    -0.15
    cies
    -0.15
    quests
    -0.14
    STYPE
    -0.14
     Lair
    -0.14
     accommod
    -0.14
     bá»ĵi
    -0.14
    ylum
    -0.13
    POSITIVE LOGITS
     stop
    0.23
     block
    0.23
     effectively
    0.23
     effective
    0.23
     stopping
    0.22
     Stop
    0.21
     cur
    0.21
     Effective
    0.21
     against
    0.21
     defeating
    0.20
    Activations Density 0.456%

    No Known Activations