Home - News ( United States | United Kingdom | Italy | Germany ) - Football scores

Showing content from https://github.com/Peterande/D-FINE/issues/169 below:

Finetuning is not stable · Issue #169 · Peterande/D-FINE · GitHub

Finetuned the best model (object365+coco) on a custom dataset:

• Dataset size: 28, 000 (for training) and 5, 000 (for validation)
• No.of classes: 18
• No.of GPUs used: 4
• No.of epochs: 80
• Batch-size: 4

Observations:

• After 10 epochs, training loss were not reduced. it was slightly decreased

Have you guys have experienced this ?
I would like to know your insights on this finetuning process!

Thank you!

RetroSearch is an open source project built by @garambo | Open a GitHub Issue

Search and Browse the WWW like it's 1997 | Search results from DuckDuckGo

HTML: 3.2 | Encoding: UTF-8 | Version: 0.7.4