Comments (2)
你好
- 训练代码 里使用了
transformers
的Trainer
和TrainingArguments
类,是支持将train和eval的过程进行记录的,具体为TrainingArguments
的report_to
这个入参,目前默认设置就是tensorboard。输出地址为output_dir/runs
,tensorboard的记录分为train和eval两个阶段。
- train对应metric有3类,如下:
- eval对应的metric有8类,包括:Roug-1、Rouge-2、Rouge-L、Rouge-Lsum,如下:
你提到的Rouge类指标是eval阶段计算的,目前默认策略是每个epoch结束后进行eval,对应TrainingArguments
的evaluation_strategy="epoch"
。如果想提高eval的频次,可修改TrainingArguments
的evaluation_strategy="steps"
,表示每训练固定步数进行一次eval,同时也需要设置对应的步长,即TrainingArguments
的eval_steps
。这2个参数为训练代码的入参,可以直接修改,如下:
python src/train_sft.py \
--evaluation_strategy="steps" \
--eval_steps 10 \
...
- 这里5.4M指的是数据量,不是文件的大小,实际训练数据文件大小就是5G
from rlhf.
from rlhf.
Related Issues (20)
- deepspeed速度
- 请教一下有考虑加IDEA-CCNL/Wenzhong2.0-GPT2-3.5B-chinese这个模型做评测比较吗 HOT 1
- No module named 'transformers_modules.sunzeyeah.pangu-2' HOT 4
- 使用LoRA的GLM-10B-chinese模型是如何保存的 HOT 3
- 用chatGLM-6B训RW的时候loss不收敛 HOT 7
- 用GLM-10B-chinese训练RLHF过程,有没有模型并行的方式? HOT 4
- reward model的实现问题 HOT 1
- rlhf deepspeed和trlx能否支持 sft chatglm 6b HOT 13
- RLHF第三步进行训练时,在modeling_glm.py如下代码位置会显示“size mismatch, got 1028, 1028x1024,0” HOT 4
- train_rlhf-trlx.py代码问题 HOT 3
- reward推理问题 HOT 5
- 关于取最后一个token作为reward分数的方式 HOT 2
- chatglm+RLHF HOT 1
- 请问支持lora方式吗 HOT 1
- 请问为什么我在基于Chatglm-6b模型执行train_sft.sh训练时,启用deepspeed没有效果 HOT 2
- 后续会考虑RLHF平替方案的集成么 HOT 2
- 有这个推理的代码吗
- 基于ChatGLM2的RLHF训练问题 HOT 1
- pretrain_data_v1.jsonl 文件在哪里 HOT 5
- Pangu 2.6b 启动失败。 HOT 1
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 rlhf.