-
Hello! I have some minor questions about certain details in the training of the network itself.
1. How are the results of the paper acquired?
From the paper, its said that:
`Following [3, 10, 1…
-
A major feature enhancement of this guide would be identifying where particular models are better suited to particular stages / or types of work.
-
DeepSpeed and Huggingface appears to be slowing training down significantly. We should investigate why -- it may be the optimizer states.
-
So, I don't know if you care enough to update this guide or not in any sense but I've sorta taken on this challenge with the more modern LFS build system and pacman package. The first biggest change. …
-
## Current
The tutorial introduces a architecture diagram and a high level sequence diagram. Afterwards it links to a developer guide.
## Problem
Students may be overwhelmed by the information in…
-
The MISMIP+ demo requires very high resolution near the grounding line and probably also the shear margins. At coarser resolutions the simulation exhibits weird oscillatory artifacts and obvious mesh …
-
Hello, thanks for your repository, it inspires me a lot. For the QA dataset, I wonder how to adapt this framework to the GrailQA dataset? What changes do I need to make to this framework?
-
你好,使用released的deepseek-math-7b-rl-stepdpo模型,在dpo10k的数据上进行推理,效果却很差,是什么原因?
按照提供的推理参数和deepseek-math对应的template
temperature=0.9,topp=0.95
prompt输入:
```json
{"role": "user", "content": "Jenny has …
-
### What should we add?
The code below contains a fairly subtle bug, and the plot it generates is not the intended plot. (Comment out the "incorrect line" and uncomment out the "correct line" to get …
-
Ive made some prototype i would like you to take a look at this , currently as a gist , ill be polishing it of course.
https://gist.github.com/rubenCodeforges/bd876bd40343d0f68616