Comments (4)
你好,
我们使用的命令是一致的,运行时不需要设置特定的超参数。
可以确认下是否由下面两个原因引起:
- 下载的wizardcoder-python-7b模型是否正确?如果按照这个repository的代码复现性能低,可以尝试下通过WizardLM中的说明进行复现,看是否会发生性能上的变化;
- 检查下使用的vllm版本,我使用的是0.1.4,不确定是否会因为使用不同的版本而影响模型inference的性能。
from mergelm.
我检查了下是否是模型问题或者vllm版本问题,发现似乎并不是这两个问题导致的。然后我对比了下wizardlm官方仓库和您的仓库推理human_eval任务的区别(wizardlm官方仓库似乎只提供了wizardcoder-python-34b的eval脚本),发现有两点不一样:
- temperature; wizardlm仓库设置是0.2,mergelm设置的是0.0
- decoding_style; wizardlm仓库设置是loops=100,mergelm是loops=1
目前我正在测试wizardlm仓库的精度,但我想问下是否可能和上面两个参数相关?
from mergelm.
我在运行时,temperature为0.0保证输出使用确定性的greedy策略,loops设置为1也不影响模型效果。感觉应该也不是两个参数的问题。
请问你有尝试WizardCoder-Python-13B或者WizardCoder-Python-34B吗?他们的性能是否可以复现?
from mergelm.
调整vllm版本为0.1.4,transformer版本为4.33.1后精度正常。感谢。
from mergelm.
Related Issues (20)
- 看起来参数是随机丢弃了 HOT 2
- Does this work with Encoder-Decoder models like T5? HOT 2
- WizardMath model embedding层维度问题。 HOT 2
- The question about encoder-based model merge. HOT 2
- 为什么融合wizard-lm和math后模型生成乱码 HOT 4
- 你好,关于融合code模型的选择问题 HOT 1
- 使用ties和magnitude方法遇到了一些问题 HOT 1
- WizardMath-7b和WizardLM-7b模型合并问题 HOT 2
- PEFT integration of DARE method HOT 1
- 使用merge_llms_instruct_math_code.py在评估Math数据集的时候CUDA out of memory HOT 2
- 如何对齐论文中 LM&Math&Code融合的指标 HOT 4
- 模型支持 HOT 2
- Couldn't find a dataset script at /home/dell7960/PycharmProjects/DARE/MergeLM/glue/glue.py or any data file in the same directory. HOT 1
- ValueError: BuilderConfig 'rte' not found. Available: ['default'] HOT 2
- AssertionError: cannot find file trainer_state.json! HOT 10
- Questions about randomly set delta parameters==zero HOT 2
- Seeking mirrors of WizardLM models HOT 3
- What is the LICENSE type of this repo? HOT 2
- Is it possible to merge 2 LLAMA derive models like LLaVA and CodeLLAMA? HOT 2
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 mergelm.