[Patch] Remove the patch of MiniCPM

[gcanlin]

What this PR does / why we need it?

Part of #5304.

After vllm-project/vllm#32523 merge, we could remove the patch of MiniCPMAttention.

Does this PR introduce any user-facing change?

How was this patch tested?

Test it locally.

from transformers import AutoTokenizer
from vllm import LLM, SamplingParams

model_name = "openbmb/MiniCPM-2B-sft-bf16"
prompt = [{"role": "user", "content": "Write an article about Artificial Intelligence."}]

tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
input_text = tokenizer.apply_chat_template(prompt, tokenize=False, add_generation_prompt=True)

llm = LLM(
    model=model_name,
    trust_remote_code=True,
    max_num_batched_tokens=65536,
    dtype="bfloat16", 
    gpu_memory_utilization=0.8, 
)
sampling_params = SamplingParams(top_p=0.95, temperature=0.6, max_tokens=32768)

outputs = llm.generate(prompts=input_text, sampling_params=sampling_params)

print(outputs[0].outputs[0].text)

INFO 01-17 15:58:34 [llm.py:347] Supported tasks: ['generate']
Adding requests: 100%|██████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00, 57.79it/s]
Processed prompts:   0%|                                            | 0/1 [00:00<?, ?it/s, est. speed input: 0.00 toks/s, output: 0.00 toks/s](EngineCore_DP0 pid=536888) INFO 01-17 15:58:34 [acl_graph.py:188] Replaying aclgraph
Processed prompts: 100%|███████████████████████████████████| 1/1 [00:08<00:00,  8.41s/it, est. speed input: 1.66 toks/s, output: 51.36 toks/s]
 Artificial Intelligence (AI) is a field of computer science that focuses on creating machines that can perform tasks that typically require human intelligence. AI has been around for decades, but it is only in recent years that it has become more accessible and practical for everyday use.

One of the most significant developments in AI is machine learning. Machine learning is a subset of AI that focuses on teaching machines to learn from data. This means that instead of being programmed to perform a specific task, machines can learn on their own by analyzing data and making decisions based on that data.

The potential applications of AI are vast and varied. In healthcare, AI can be used to analyze medical images and help doctors diagnose diseases. In finance, AI can be used to analyze financial data and predict market trends. In transportation, AI can be used to optimize traffic flow and reduce congestion.

AI is also being used in gaming and entertainment. Games like AlphaGo have demonstrated the potential of AI to outperform humans at complex games like Go. AI is also being used in movies and TV shows to create realistic characters and environments.

While AI has many potential applications, there are also concerns about its impact on society. One concern is the potential loss of jobs as machines become more capable of performing tasks traditionally done by humans. There are also concerns about the ethical implications of AI, such as the use of facial recognition technology to identify and track individuals without their consent.

Despite these concerns, the potential benefits of AI are significant. AI has the potential to revolutionize many industries and improve our lives in countless ways. As AI continues to develop, it will be important for policymakers and industry leaders to address these concerns and ensure that AI is used in a responsible and ethical manner.

In conclusion, AI is a field of computer science that has the potential to revolutionize many industries and improve our lives in countless ways. While there are concerns about its impact on society, the potential benefits of AI are significant. As AI continues to develop, it will be important for policymakers and industry leaders to address these concerns and ensure that AI is used in a responsible and ethical manner.

• vLLM version: v0.13.0
• vLLM main: vllm-project/vllm@2c24bc6

[github-actions]

👋 Hi! Thank you for contributing to the vLLM Ascend project. The following points will speed up your PR merge:‌‌

• A PR should do only one thing, smaller PRs enable faster reviews.
• Every PR should include unit tests and end-to-end tests ‌to ensure it works and is not broken by other future PRs.
• Write the commit message by fulfilling the PR description to help reviewer and future developers understand.

If CI fails, you can run linting and testing checks locally according Contributing and Testing.

[gemini-code-assist]

Code Review

This pull request removes an obsolete patch for MiniCPMAttention, which is a good cleanup as it simplifies the codebase. However, the corresponding test file tests/ut/patch/worker/patch_common/test_patch_minicpm.py has not been removed. This will cause an ImportError and break the test suite. Please remove the test file as well.

I am having trouble creating individual review comments. Click here to see my feedback.

vllm_ascend/patch/worker/patch_minicpm.py (18-36)

With the removal of this patch file, the corresponding test file tests/ut/patch/worker/patch_common/test_patch_minicpm.py should also be removed. The test file imports from this patch file, and its removal without deleting the test file will lead to an ImportError, causing the CI to fail.

[gcanlin]

Waiting for main branch upgrading to 0118. Then we can merge this PR.

[wangxiyuan]

let's block this change until the vllm change works with vllm ascend on main

[gcanlin]

@wangxiyuan Ready to merge now.

[wangxiyuan]

no, the vllm commit vllm-project/vllm@fe36bf5 is not included in v0.14.1. Let's merge this once we upgrade to 0.15

[gcanlin]

no, the vllm commit vllm-project/vllm@fe36bf5 is not included in v0.14.1. Let's merge this once we upgrade to 0.15

Oh, okay. We're still keeping compatibility with 0.14.1.

[github-actions]

This pull request has conflicts, please resolve those before we can evaluate the pull request.

[wangxiyuan]

this PR can be rebased and merged now. @gcanlin

[gcanlin]

this PR can be rebased and merged now. @gcanlin

Done.