未验证 提交 1c321287 编写于 作者: W Walter 提交者: GitHub

Merge pull request #2071 from RainFrost1/dynamic_benchmark

add r50 dynamic fp16 train for benchmark
...@@ -51,6 +51,6 @@ inference:python/predict_cls.py -c configs/inference_cls.yaml ...@@ -51,6 +51,6 @@ inference:python/predict_cls.py -c configs/inference_cls.yaml
null:null null:null
null:null null:null
===========================train_benchmark_params========================== ===========================train_benchmark_params==========================
batch_size:128 batch_size:128|256
fp_items:amp_fp16 fp_items:ampfp16
epoch:1 epoch:1
...@@ -51,6 +51,6 @@ inference:python/predict_cls.py -c configs/inference_cls.yaml ...@@ -51,6 +51,6 @@ inference:python/predict_cls.py -c configs/inference_cls.yaml
null:null null:null
null:null null:null
===========================train_benchmark_params========================== ===========================train_benchmark_params==========================
batch_size:128 batch_size:128|256
fp_items:pure_fp16 fp_items:purefp16
epoch:1 epoch:1
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册