Jingcheng Hu

Please refer to README.md in each directory. Student Name: Jingcheng Hu Student ID: 2022312848 Contact: [email protected] or 18170915210

Is the data collection prompt for detailed decription and complex reasoning released?

1

Many thanks for your great work! I am loving the idea of using text-only-gpt-4 to harness img-based multimodal data. And I am checking your paper on Table 10 about the...

Really appreciate your great work! And I am especially interesting in the part of how you will collect the comparison data and how you guys maintain the quality of the...

RLHF training code for StableVicuna open sourced?

1

Very exciting to see you guys' remarkable work on stablevicuna!! And I read through your blog and notice that all the dataset is open sourced and available; however, considering the...

Many thanks for your great open sourcing effort! And I am new to this field and I am particularly interested in the data for training reward model. I noticed that...

[QUESTION] Why megatron-core seems slower and use more gpu mem than legacy for gpt_pretrain?

1

**Your question** I run pretrain_gpt on same arch, data, training hyperparams and same hardware, with and without using megatron_core when build the model. I notice clearly **worse wall clock time...

Release examples of training stablevicuna

3

### 🚀 The feature, motivation, and pitch Very exciting to see you guys' remarkable effort for open source this repo! And I read through stablevicuna blog and notice that the...