Could you please tell me the commands step by step to run your model? Thank you!

[Ricardokevins]

No description provided.

Hi, I thnk the process should be as followed:

1. pip install -r requirement.txt to install the env.
2. Choose a model you like to download from here: MAPO-collection
3. Step into Evaluation and run the script file: run.sh (model path should be set correctly)

If you have any question, feel free to raise the issue.

[yyyhz]

No description provided. 未提供说明。

Hi, I thnk the process should be as followed:嗨，我该过程应如下：

1. pip install -r requirement.txt to install the env.pip install -r requirement.txt安装环境。
2. Choose a model you like to download from here: MAPO-collection从这里选择您喜欢下载的模型：MAPO-collection
3. Step into Evaluation and run the script file: run.sh (model path should be set correctly)单步执行评估并运行脚本文件：run.sh（模型路径应正确设置）

If you have any question, feel free to raise the issue.如果您有任何问题，请随时提出问题。

Thank you for your reply! But I would like to reproduce your code, e.g. step-by-step from MathOctopus to MathOctopus-MAPO. In the reproduction process, there may be some models missing (including but not limited to /mnt/data/shesj/PLM/nllb-200-distilled-600M, etc) as well as more difficult to modify code(including but not limited to No module named 'utils.generatePrompt'). not sure. I would be appreciated if you can focus on this.

[Ricardokevins]

No description provided. 未提供说明。

Hi, I thnk the process should be as followed:嗨，我该过程应如下：

1. pip install -r requirement.txt to install the env.pip install -r requirement.txt安装环境。
2. Choose a model you like to download from here: MAPO-collection从这里选择您喜欢下载的模型：MAPO-collection
3. Step into Evaluation and run the script file: run.sh (model path should be set correctly)单步执行评估并运行脚本文件：run.sh（模型路径应正确设置）

If you have any question, feel free to raise the issue.如果您有任何问题，请随时提出问题。

Thank you for your reply! But I would like to reproduce your code, e.g. step-by-step from MathOctopus to MathOctopus-MAPO. In the reproduction process, there may be some models missing (including but not limited to /mnt/data/shesj/PLM/nllb-200-distilled-600M, etc) as well as more difficult to modify code(including but not limited to No module named 'utils.generatePrompt'). not sure. I would be appreciated if you can focus on this.

Thank you for your suggestion! I am working on providing a more detailed instructions

[yyyhz]

No description provided. 未提供说明。未提供说明。未提供说明。

Hi, I thnk the process should be as followed:嗨，我该过程应如下：嗨，我该过程应如下：嗨，我该过程应如下：

1. pip install -r requirement.txt to install the env.pip install -r requirement.txt安装环境。pip install -r requirement.txt安装 env.pip install -r requirement.txt安装环境。
2. Choose a model you like to download from here: MAPO-collection从这里选择您喜欢下载的模型：MAPO-collection从这里选择您喜欢下载的模型： MAPO-collection从这里选择您喜欢下载的模型：MAPO-collection
3. Step into Evaluation and run the script file: run.sh (model path should be set correctly)单步执行评估并运行脚本文件：run.sh（模型路径应正确设置）

If you have any question, feel free to raise the issue.如果您有任何问题，请随时提出问题。

Thank you for your reply! But I would like to reproduce your code, e.g. step-by-step from MathOctopus to MathOctopus-MAPO. In the reproduction process, there may be some models missing (including but not limited to /mnt/data/shesj/PLM/nllb-200-distilled-600M, etc) as well as more difficult to modify code(including but not limited to No module named 'utils.generatePrompt'). not sure. I would be appreciated if you can focus on this.感谢您的回复！但我想重现您的代码，例如从MathOctopus逐步到MathOctopus-MAPO。在复制过程中，可能会缺少一些模型（包括但不限于 /mnt/data/shesj/PLM/nllb-200-distilled-600M 等）以及更难修改的代码（包括但不限于 No module named 'utils.generatePrompt'）。不确定。如果您能专注于此，我将不胜感激。

Thank you for your suggestion! I am working on providing a more detailed instructions谢谢你的建议！我正在努力提供更详细的说明

And I have a more urgent question, I didn't find your open source utils.py file. This is a very urgent issue for me and I would appreciate it if you could open source it! No module named 'utils.generatePrompt'

[Ricardokevins]

I have tried to fix the problem

You can pull the latest code and try again

[yyyhz]

I have tried to fix the problem我已尝试解决问题

You can pull the latest code and try again您可以提取最新代码，然后重试

Thanks for the update! If it's possible, could you please provide the contents of your mnt (e.g. /mnt/data/shesj/PLM/nllb-200-distilled-600M, /mnt/data/shesj/RL4CoT, etc.)

[Ricardokevins]

I have tried to fix the problem我已尝试解决问题 You can pull the latest code and try again您可以提取最新代码，然后重试

Thanks for the update! If it's possible, could you please provide the contents of your mnt (e.g. /mnt/data/shesj/PLM/nllb-200-distilled-600M, /mnt/data/shesj/RL4CoT, etc.)

The original path for my algorithm development, /mnt/data/shesj/RL4CoT, contains a large number of disorganized files. The code in the repository has been refactored and reorganized.

nllb-200-distilled-600M is a public translation model, you can download it from huggingface and you can also refer to the paper for more details.

[yyyhz]

I have tried to fix the problem我已尝试解决问题 You can pull the latest code and try again您可以提取最新代码，然后重试

Thanks for the update! If it's possible, could you please provide the contents of your mnt (e.g. /mnt/data/shesj/PLM/nllb-200-distilled-600M, /mnt/data/shesj/RL4CoT, etc.)

The original path for my algorithm development, /mnt/data/shesj/RL4CoT, contains a large number of disorganized files. The code in the repository has been refactored and reorganized.

nllb-200-distilled-600M is a public translation model, you can download it from huggingface and you can also refer to the paper for more details.

I have used your lastest code, it is work for me!

1. But I have new problem when I try to run ppo.py. Change the model in ppo_trainer to _model.criticmodel and there will be no error. But I don't know if that's correct.
2. And also I am not sure where is the explore_data in this PPO config file comes from, the readme doesn't show.

[Ricardokevins]

I have tried to fix the problem我已尝试解决问题 You can pull the latest code and try again您可以提取最新代码，然后重试

Thanks for the update! If it's possible, could you please provide the contents of your mnt (e.g. /mnt/data/shesj/PLM/nllb-200-distilled-600M, /mnt/data/shesj/RL4CoT, etc.)

The original path for my algorithm development, /mnt/data/shesj/RL4CoT, contains a large number of disorganized files. The code in the repository has been refactored and reorganized. nllb-200-distilled-600M is a public translation model, you can download it from huggingface and you can also refer to the paper for more details.

I have used your lastest code, it is work for me!

1. But I have new problem when I try to run ppo.py. Change the model in ppo_trainer to _model.criticmodel and there will be no error. But I don't know if that's correct.
2. And also I am not sure where is the explore_data in this PPO config file comes from, the readme doesn't show.

Due to changes in the working environment, there might be some inconsistency in some file names. I have made many effort to locate the previous data files, and you can refer to the files in the following link for use.

https://github.com/NJUNLP/MAPO/blob/main/Data/numglue-mutli-lingual_NoEnglishForPPO.json