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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Qwen3-32B
    5. 32-RESID-BATCHTOPK-65K
    7. 5196
    Prev
    Next
    INDEX
    Explanations

    forgetting information

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    adamkarvonen/qwen3-32b-saes/saes_Qwen_Qwen3-32B_batch_top_k/resid_post_layer_32
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    èµĮ
    -0.10
    åĭĸ
    -0.09
    åıĮéĩį
    -0.08
     jal
    -0.08
    opies
    -0.08
    .dependencies
    -0.08
    è£Ļ
    -0.08
    为ä¸Ģä½ĵ
    -0.08
    æĦį
    -0.08
    arty
    -0.08
    POSITIVE LOGITS
    éģĹå¿ĺ
    0.20
    è®°å¿Ĩ
    0.17
    çļĦè®°å¿Ĩ
    0.16
    å¿ĺ
    0.15
     forget
    0.14
    å¿ĺè®°äºĨ
    0.14
     memory
    0.13
    å¿ĺäºĨ
    0.13
    æµģ失
    0.13
    å¿ĺè®°
    0.13
    Activations Density 0.063%

    No Known Activations