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

    Neuronpedia

    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. 25-LLAMASCOPE-RES-32K
    11. 11463
    Prev
    Next
    INDEX
    Explanations

    references to ancient civilizations and historical contexts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With LLAMA3.1-8B @ 25-llamascope-res-32k
    Configuration
    fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L25R-8x
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    cerebras/SlimPajama-627B
    Features
    32,768
    Data Type
    bfloat16
    Hook Name
    blocks.25.hook_resid_post
    Hook Layer
    25
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    cerebras/SlimPajama-627B
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    xmax
    -0.14
    angep
    -0.14
    Benchmark
    -0.13
    оки
    -0.13
    prise
    -0.13
    glfw
    -0.13
    dsl
    -0.13
     assertNotNull
    -0.13
    à¸ł
    -0.13
     Benchmark
    -0.12
    POSITIVE LOGITS
     ancient
    0.55
     Ancient
    0.52
    Anc
    0.46
    anc
    0.45
     anc
    0.40
    åı¤
    0.36
     Anc
    0.33
     дÑĢев
    0.31
    ANC
    0.31
     åı¤
    0.30
    Activations Density 0.491%

    No Known Activations