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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. OpenMOSS · Llama Scope: SAEs for Llama-3.1-8B
    3. Llama3.1-8B (Base)
    4. Residual Stream
    5. 21-LLAMASCOPE-RES-32K
    6. 9623
    Prev
    Next
    INDEX
    Explanations

    phrases expressing fluctuations or changes, particularly those that indicate positive (ups) and negative (downs) experiences

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With LLAMA3.1-8B @ 21-llamascope-res-32k
    Configuration
    fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L21R-8x
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    cerebras/SlimPajama-627B
    Features
    32,768
    Data Type
    bfloat16
    Hook Name
    blocks.21.hook_resid_post
    Hook Layer
    21
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    cerebras/SlimPajama-627B
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    lessly
    -0.17
    296
    -0.15
    cken
    -0.15
     deepest
    -0.15
     Erk
    -0.15
     Pant
    -0.15
    ovice
    -0.14
    opsis
    -0.14
    ort
    -0.14
     Grip
    -0.14
    POSITIVE LOGITS
    /down
    0.44
    -down
    0.25
    /up
    0.23
    datable
    0.20
    ilon
    0.20
    scaling
    0.19
    graded
    0.19
    most
    0.19
    ward
    0.19
    ILON
    0.18
    Activations Density 0.049%

    No Known Activations