About pre-training issues #23

Activity

1. LMMMEng commented on Dec 31, 2025

   

   LMMMEng

   on Dec 31, 2025 via email

   Owner

   More actions

   Hi, Happy New Year. When training on ImageNet-1K, the model is trained from scratch. For downstream tasks (segmentation and detection), the model utilizes the pre-trained weights obtained from ImageNet. ---- Replied Message ---- ***@***.***>Date12/31/2025 ***@***.***>***@***.***>Subject[LMMMEng/TransXNet] About pre-training issues (Issue #23) Fimwu7 created an issue (LMMMEng/TransXNet#23) Hello Author, Happy New Year. I would like to ask, what weights were used for pre-training your model? I look forward to your reply. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>