[doc] Add Qwen2.5 tutorials #4636
Conversation
Model module is useless now. Let't remove it totally. - vLLM version: v0.11.2 Signed-off-by: yangshihao6 <[email protected]>
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request adds a new tutorial for deploying the Qwen2.5-7B-Instruct model. The tutorial is comprehensive but appears to be an internal document that has been published without being adapted for a public audience. It contains several internal links, references to internal Docker images, and network-specific instructions that make it impossible for external users to follow. My review focuses on highlighting these issues to ensure the documentation is accessible and reproducible for the open-source community.
docs/source/tutorials/Qwen2.5-7b.md
Outdated
| -v /usr/local/Ascend/driver:/usr/local/Ascend/driver \ | ||
| -v /home/:/home \ | ||
| -w /home/<name> \ | ||
| mindie:dev-2.2.RC1.B070-800I-A2-py312-ubuntu22.03-x86_64 \ |
There was a problem hiding this comment.
The Docker image mindie:dev-2.2.RC1.B070-800I-A2-py312-ubuntu22.03-x86_64 appears to be an internal image, which is not available to the public. For this tutorial to be reproducible by the open-source community, please use a publicly available base image and provide instructions to build upon it if necessary.
docs/source/tutorials/Qwen2.5.md
Outdated
| :::{note} | ||
| The server may be in a restricted network zone (Yellow Zone) and require a Green Zone proxy tool for internet access. Download the proxy tool from the internal repository, run `PortMapping.exe` to obtain the proxy IP, and update `ip_addr` in `portproxy_remote.sh` before executing the script. | ||
| ::: |
There was a problem hiding this comment.
The note about "Yellow Zone", "Green Zone proxy tool", and internal repositories is specific to a corporate network environment. This information is not relevant or helpful for external users and makes the documentation confusing. Please remove these internal-specific instructions to make the guide generally applicable.
| Qwen2.5-7B-Instruct supports single-node single-card deployment on the 910B4 platform. Follow these steps to start the inference service: | ||
|
|
||
| 1. Prepare model weights: Ensure the downloaded model weights are stored in the `./Qwen2.5-7B-Instruct/` directory. | ||
| 2. Download the gsm8k dataset (for evaluation): [gsm8k.zip](https://vision-file-storage/api/file/download/attachment-v2/WIKI202511118986704/32978033/20251111T144846Z_9658c67a0fb349f9be081ab9ab9fd2bc.zip?attachment_id=32978033) |
There was a problem hiding this comment.
The link to the gsm8k.zip dataset points to an internal vision-file-storage URL, which is not accessible to the public. This prevents users from following the tutorial. Please replace it with a public link to the dataset, for example, from the Hugging Face Hub.
docs/source/tutorials/Qwen2.5.md
Outdated
|
|
||
| #### Evaluation Results | ||
| Results and logs are saved to `benchmark/outputs/default/`. A sample accuracy report is shown below: | ||
|  |
There was a problem hiding this comment.
The image for "Accuracy Evaluation Result" is hosted on an internal wiki.huawei.com server and is inaccessible to the public. To make the documentation useful for everyone, please embed the image directly into the repository and use a relative path. This also applies to the "Performance Evaluation Result" image on line 307.
|
👋 Hi! Thank you for contributing to the vLLM Ascend project. The following points will speed up your PR merge:
If CI fails, you can run linting and testing checks locally according Contributing and Testing. |
Model module is useless now. Let't remove it totally. - vLLM version: v0.11.2 Signed-off-by: yangshihao6 <[email protected]>
github-actions
bot
added
the
documentation
Model module is useless now. Let't remove it totally. - vLLM version: v0.11.2 Signed-off-by: yangshihao6 <[email protected]>
yangshihao6
changed the title
Model module is useless now. Let't remove it totally. - vLLM version: v0.11.2 Signed-off-by: yangshihao6 <[email protected]>
# Conflicts: # docs/source/tutorials/Qwen2.5.md
Model module is useless now. Let't remove it totally. - vLLM version: v0.11.2 Signed-off-by: yangshihao6 <[email protected]>
2 participants
What this PR does / why we need it?
Does this PR introduce any user-facing change?
How was this patch tested?