[Example] ggml: add llava support

[dm4]

• Also update the CI badges in README.md
• Merge after this related WasmEdge PR

[juntao]

Hello, I am a code review bot on flows.network. Here are my reviews of code commits in this PR.

---

Potential Issues and Errors:

1. Inconsistencies in error handling mechanisms and buffer size management in the codebase.
2. Usage of unwrap that may lead to panics rather than graceful error handling.
3. Lack of clarity and detailed error handling logic in various functions.
4. Code duplication in CI workflows and missing testing for the new llava build step.
5. Lack of information on code review process by other team members.
6. Changes across multiple files without clear description of new features or changes in functionality.
7. Recommendations in README.md may lead to confusion for users not using the specified llava versions.
8. Modification in main.rs messages and placeholders could affect user experience.
9. Lack of context or clear description for a simple text correction in the llama job configuration.

Important Findings:

• The patch introduces new functionality for llava support within the ggml project.
• It includes updating README files, adding CI badges, modifying code for llava interaction, and expanding workflows.
• Areas requiring immediate attention include improved error handling, eliminating duplications, comprehensive testing, and ensuring code review practices.
• Clear documentation and communication regarding changes are essential to avoid confusion and maintain consistency within the project.

Considering the potential issues and significant findings in the patch, it is crucial to address error handling, streamline workflows, clarify changes, and enrich the review process before merging to enhance the overall robustness and maintainability of the ggml project with llava support.

Details

Commit e36591ebe131f430253ddbeafef3d5ba55e06044

Key changes in the patch:

1. Added support for llava in the ggml project.
2. Added Cargo.toml, README.md, and main.rs files to the llava directory.
3. Modified the README.md in the ggml project to include new parameters for llava.
4. Updated the README.md in the llava directory with instructions on using the llava model.
5. Added functionality in main.rs to interact with llava model, set context data, and get output.

Potential problems:

1. In the read_input function in main.rs, the loop reading input from stdin may not handle cases where input is not provided as expected. It should have a way to handle end-of-file conditions.
2. The set_data_to_context function in main.rs does not return any error handling mechanism. It should consider returning a Result to handle potential errors.
3. The get_data_from_context function in main.rs may have a potential issue with buffer size management. It assumes a fixed buffer size for output without dynamic resizing as needed.
4. The use of unwrap in various parts of the code in main.rs could cause the program to panic if unwrapping fails. It is recommended to handle errors more gracefully.
5. When checking environment variables in main.rs, it is better to use a match statement instead of multiple if let statements for better error handling and clarity.
6. The handling of resetting context in main.rs needs further clarification and possibly more detailed error handling logic.

Overall, the changes introduce new functionality for llava support, but the code could benefit from improved error handling and robustness.

Commit 9ee94dd3bbd58c7dd00951f79f895da518d09fd2

Key Changes:

1. Fixed the CI badges in the README file by updating the links and badge images to point to different workflows (llama, pytorch, tflite).
2. Added three new CI badges for the llama, pytorch, and tflite workflows.
3. Removed the previous CI badge that was pointing to a specific workflow (build.yaml).

Potential Problems:

1. The new CI badges added for the llama, pytorch, and tflite workflows might not be correctly configured. It would be essential to verify that these workflows are correctly set up and functioning as intended.
2. The removal of the previous CI badge may lead to confusion for users who were familiar with that specific CI workflow. It might be worth considering adding a note in the README explaining the changes made to the CI badges and workflows.

Commit 155ef14ffeae8a6a1208ca876bd9db50750b66d5

Key Changes:

1. Added a new workflow step to build "llava" in the YAML file .github/workflows/llama.yml. This step involves running a cargo build command with specific target options.
2. The new step to build "llava" is included in both the Linux and macOS job configurations.

Potential Problems:

1. Code Duplication: The same build step for "llava" is repeated for both Linux and macOS. It could be improved by extracting the common steps to avoid duplication.
2. Missing Testing: The Pull Request lacks information on whether the "llava" build step has been tested and verified for correctness and compatibility.
3. No Code Reviewer: It's not clear if the changes have been reviewed by a team member other than the author. It's essential to have a second pair of eyes to catch potential errors.
4. Limited Scope: The provided patch is focused on adding llava support, but it would be beneficial to include more contextual information about the purpose of llava and its integration within the project.

It's recommended to address these points and ensure a comprehensive review process before merging the changes.

Commit d7fe7b2282a24618bb3a8f889f45d655e9d7983f

Key Changes:

1. Added README files for chatml, llama-stream, and llama.
2. Updated the sizes of the respective .wasm files and modified the content within these files.
3. Modified the main.rs file in chatml to change some print statements.
4. Various changes to the README, main.rs, and .wasm files in the mentioned directories.

Potential Problems:

1. The patch introduces changes to multiple files across different directories. It may be challenging to review the impact of these changes without viewing the full context of the project.
2. The addition of new features or changes in functionality has not been clearly described in the patch.
3. The modification of the .wasm files should be reviewed to ensure that the changes are correct and do not introduce potential bugs.
4. The consistency of code style and formatting should be maintained across all the changes made in the patch.

Commit 1614383b11b25229005bdaadaeb5ea6811381659

Key Changes:

1. Updated the llava example in the README.md with recommendations on using ctx-size based on the llava version.
2. Modified the main.rs file to adjust the printed messages and placeholders in the llava chat format.
3. The llava.wasm file was updated, resulting in a slight reduction in size.

Potential Problems:

1. The update in the README.md recommends using different values for ctx-size based on different llava versions (llava-v1.5 and llava-v1.6). This change may lead to confusion or inconsistencies if users are not using the recommended versions or if the recommendations are not clearly stated in the documentation.
2. In the main.rs file, the change in messages and placeholders may impact the user experience or understanding of the llava chat format. It's important to ensure that these changes do not disrupt the functionality or usage of the llava application.
3. The binary patch in the llava.wasm file might be difficult to review or understand without the full context of the changes made. It's essential to provide clear descriptions or details of the changes in binary files for better understanding and review.

These potential problems should be further analyzed and addressed to ensure the effective implementation and usability of the llava support in the project.

Commit b68249bc2e89e441185cf7fc0798c7ea5e7e474a

Key Changes:

1. The patch fixes a typo in the llama job configuration in the .github/workflows/llama.yml file.
2. Specifically, the change updates the response of the AI assistant in the command that the llama job runs.

Potential Problems:

1. The change appears to be a simple text correction, so there shouldn't be high risks associated with it.
2. However, there is a lack of context about why this change was made. It would be beneficial to have a clear description of the purpose of this adjustment in the commit message for better understanding.
3. It might be a good idea to review the overall workflow file to ensure that there are no further errors or inconsistencies present, especially in job configurations.

Overall, this patch seems straightforward and does not introduce significant issues. It would be advisable to have more detailed commit messages and a broader review of the workflow file for thoroughness.