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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Julius Han · Gated SAE for Llama3-8B-Instruct
    5. Llama3-8B-IT
    7. Residual Stream
    9. 25-RES-JH
    11. 63356
    Prev
    Next
    INDEX
    Explanations

    logical inconsistencies or contradictions in arguments

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    Juliushanhanhan/llama-3-8b-it-res/blocks.25.hook_resid_post
    Features
    65,536
    Data Type
    float32
    Hook Name
    blocks.25.hook_resid_post
    Hook Layer
    25
    Architecture
    gated
    Context Size
    1,024
    Dataset
    Juliushanhanhan/openwebtext-1b-llama3-tokenized-cxt-1024
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    latable
    -0.15
    aman
    -0.15
    okus
    -0.15
    'options
    -0.14
    .tm
    -0.14
    ugin
    -0.14
    unny
    -0.14
    èĻ
    -0.13
    allenge
    -0.13
    arding
    -0.13
    POSITIVE LOGITS
     reasoning
    0.72
     logic
    0.71
     thinking
    0.66
     Logic
    0.62
    logic
    0.61
    Logic
    0.55
    thinking
    0.53
     Thinking
    0.52
    Thinking
    0.49
    _logic
    0.49
    Activations Density 0.027%

    No Known Activations