Comments (13)
加速与否是learning-rate决定的
from bert-gpu.
num_train_steps=10000,3卡,实际30000
from bert-gpu.
README有写
from bert-gpu.
README有写
请问这里one data LR fixed和one batch LR fixed分别是什么意思呢?
我有一个疑问,我数据总量100条数据,在3张卡跑num_train_steps=100,你的意思是我这一百条数据分别在每张卡跑100step吗?这样同等数据下3*100=300steps?是这个意思吗?
from bert-gpu.
300条数据
from bert-gpu.
字面意思
from bert-gpu.
300条数据
我的数据总量只有100条,你说的300条数据是什么意思?
from bert-gpu.
相当于300
from bert-gpu.
相当于300
行,麻烦再问一个问题:
你这个多GPU和普通的不太一样啊。100条数据不是给3张卡一起处理,而是扩成300条给3张卡处理。
如果通过增大LR来减少训练时间的话,那域训练和微调时候性能是不是必然会下降啊?按照你这个代码的逻辑
from bert-gpu.
没人逼你用我这个
from bert-gpu.
from bert-gpu.
没人逼你用我这个
抱歉,是我言语有些激烈了。但是这问题确实想知道您的一些见解
100条数据3张卡,扩成300条给3张卡处理。
如果通过增大LR来减少训练时间的话,那域训练和微调时候性能是不是必然会下降啊?不知道我哪里有误解
from bert-gpu.
learning rate 变大是 batch size 变大带来的,每个data产生的反向传播梯度不变,所以效果不下降
from bert-gpu.
Related Issues (20)
- run_pretraining_gpu.py not working HOT 5
- wrong when run_pretraining_gpu_v2 with init_checkpoint HOT 3
- Output model files compatible with Official Bert's pre-trained models? HOT 9
- OOM error HOT 1
- Suffer the Error: tensorflow.python.framework.errors_impl.InvalidArgumentError HOT 8
- Question about "init_checkpoint" and "output_dir" checkpint HOT 9
- XLNet support HOT 3
- GPT support
- The `global_step` update in `optimization_gpu.py` (line 74-75) is redundant. HOT 2
- TensorFlow2 support
- num_train_steps是一块卡还是多块卡的step? HOT 4
- 《How To Pre-train BERT In GPUs》
- 【Try】1-GPU pretrain with big learning rate for 100W-step, then 1-GPU pretrain with small learning rate for another 100W-step.
- train_batch_size and time required to pretrain HOT 2
- train 10W steps结束后,do_eva阶段出现错误 HOT 3
- Should line 74-75 in optimization_gpu.py be comment out? HOT 1
- an error like this : Segmentation fault (core dumped),Is the configuration wrong? HOT 9
- model_fn should return an EstimatorSpec. HOT 5
- 请问如何进一步实现梯度累计的功能?
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from bert-gpu.