© 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. Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
    5. Llama3.1-8B-IT (Instruct)
    7. Resid Post - 131k
    9. 11-RESID-POST-AA
    11. 6597
    Prev
    Next
    INDEX
    Explanations

    not if

    np_max-act · gemini-2.0-flash

    Clause 7 type: Amendment Right. Clause 9 type: Acceptance by Conduct (Consent to Terms). Clause 24 type: Third-Party Disclaimer. Clause 57 type: Content Moderation Discretion. Clause 74 type: Pricing and Charges Reservation. Clause 95 type: Exclusion of Consequential Damages. Clause 96 type: General Liability Exclusion. Clause 97 type: Force Majeure / Excuse for Failure to Perform. Clause 99 type: Aggregate Liability Cap. Clause 100 type: Third-Party Service Disclaimer. Clause 104 type: Governing Law. Explanation of neuron 4 behavior: the main thing this neuron does is find numbered contractual clauses.

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_11/trainer_1
    Dataset (Dashboard)
    Various
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.11.hook_resid_post
    Architecture
    standard
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     occult
    -0.07
    Price
    -0.06
     grill
    -0.06
     čt
    -0.06
    .Find
    -0.06
    ANDOM
    -0.06
     stiff
    -0.06
     immortal
    -0.06
    '↵
    -0.06
    #+
    -0.06
    POSITIVE LOGITS
    τ
    0.07
     인터
    0.07
     �
    0.07
    GameObject
    0.07
     parameter
    0.06
    horse
    0.06
    ované
    0.06
    ugeot
    0.06
    ّل
    0.06
    ter
    0.06
    Activations Density 0.002%

    No Known Activations