© 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. 2-LLAMASCOPE-RES-32K
    11. 21072
    Prev
    Next
    INDEX
    Explanations

    instances of the word "did" in various forms, indicating references to actions not taken

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With LLAMA3.1-8B @ 2-llamascope-res-32k
    Configuration
    fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L2R-8x
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    cerebras/SlimPajama-627B
    Features
    32,768
    Data Type
    bfloat16
    Hook Name
    blocks.2.hook_resid_post
    Hook Layer
    2
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    cerebras/SlimPajama-627B
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    259
    -0.16
    278
    -0.14
    dyby
    -0.14
    439
    -0.14
    APTER
    -0.13
    alon
    -0.13
    enal
    -0.13
     ****************************************************************************
    -0.13
    ileo
    -0.13
    ubar
    -0.13
    POSITIVE LOGITS
     èĥ
    0.14
    heit
    0.14
    оÑıн
    0.13
    mana
    0.13
    andel
    0.13
     Booker
    0.13
    .addAll
    0.13
    ajo
    0.12
    aji
    0.12
    жÑĥ
    0.12
    Activations Density 0.213%

    No Known Activations