© 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. 17-LLAMASCOPE-RES-32K
    11. 24994
    Prev
    Next
    INDEX
    Explanations

    instances of deception and hidden identities

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With LLAMA3.1-8B @ 17-llamascope-res-32k
    Configuration
    fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L17R-8x
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    cerebras/SlimPajama-627B
    Features
    32,768
    Data Type
    bfloat16
    Hook Name
    blocks.17.hook_resid_post
    Hook Layer
    17
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    cerebras/SlimPajama-627B
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ->__
    -0.15
    ntag
    -0.14
     Cutting
    -0.13
    ãĥ¼ãĤ¿ãĥ¼
    -0.13
    нÑĤ
    -0.13
    bias
    -0.13
    abox
    -0.13
    ighbor
    -0.13
    ixer
    -0.13
    orsche
    -0.13
    POSITIVE LOGITS
     fake
    0.28
     Fake
    0.28
     faker
    0.24
    fake
    0.23
    Fake
    0.22
     lá»
    0.21
     convincing
    0.21
     lies
    0.20
     lie
    0.20
     deception
    0.20
    Activations Density 0.252%

    No Known Activations