[Model] Adding sliding window support for block table [#3665]

[ruthe98]

I added sliding window support for the block table. I still need to do some testing and add support for the paged attention feature.

FIX #3665

---

PR Checklist (Click to Expand)

Thank you for your contribution to vLLM! Before submitting the pull request, please ensure the PR meets the following criteria. This helps vLLM maintain the code quality and improve the efficiency of the review process.

PR Title and Classification

Only specific types of PRs will be reviewed. The PR title is prefixed appropriately to indicate the type of change. Please use one of the following:

• [Bugfix] for bug fixes.
• [CI/Build] for build or continuous integration improvements.
• [Doc] for documentation fixes and improvements.
• [Model] for adding a new model or improving an existing model. Model name should appear in the title.
• [Frontend] For changes on the vLLM frontend (e.g., OpenAI API server, LLM class, etc.)
• [Kernel] for changes affecting CUDA kernels or other compute kernels.
• [Core] for changes in the core vLLM logic (e.g., LLMEngine, AsyncLLMEngine, Scheduler, etc.)
• [Hardware][Vendor] for hardware-specific changes. Vendor name should appear in the prefix (e.g., [Hardware][AMD]).
• [Misc] for PRs that do not fit the above categories. Please use this sparingly.

Note: If the PR spans more than one category, please include all relevant prefixes.

Code Quality

The PR need to meet the following code quality standards:

• We adhere to Google Python style guide and Google C++ style guide.
• Pass all linter checks. Please use format.sh to format your code.
• The code need to be well-documented to ensure future contributors can easily understand the code.
• Include sufficient tests to ensure the project to stay correct and robust. This includes both unit tests and integration tests.
• Please add documentation to docs/source/ if the PR modifies the user-facing behaviors of vLLM. It helps vLLM user understand and utilize the new features or changes.

Notes for Large Changes

Please keep the changes as concise as possible. For major architectural changes (>500 LOC excluding kernel/data/config/test), we would expect a GitHub issue (RFC) discussing the technical design and justification. Otherwise, we will tag it with rfc-required and might not go through the PR.

What to Expect for the Reviews

The goal of the vLLM team is to be a transparent reviewing machine. We would like to make the review process transparent and efficient and make sure no contributor feel confused or frustrated. However, the vLLM team is small, so we need to prioritize some PRs over others. Here is what you can expect from the review process:

• After the PR is submitted, the PR will be assigned to a reviewer. Every reviewer will pick up the PRs based on their expertise and availability.
• After the PR is assigned, the reviewer will provide status update every 2-3 days. If the PR is not reviewed within 7 days, please feel free to ping the reviewer or the vLLM team.
• After the review, the reviewer will put an action-required label on the PR if there are changes required. The contributor should address the comments and ping the reviewer to re-review the PR.
• Please respond to all comments within a reasonable time frame. If a comment isn't clear or you disagree with a suggestion, feel free to ask for clarification or discuss the suggestion.

Thank You

Finally, thank you for taking the time to read these guidelines and for your interest in contributing to vLLM. Your contributions make vLLM a great tool for everyone!

[cadedaniel]

Thanks for the contribution @ruthe98 . To close this issue I am looking for the following:

• A unit test at the block table level, e.g. at this level
• An e2e test with mistralai/Mistral-7B-v0.1 (sliding window model) that we don't crash when going beyond the sliding window
• An e2e test with the same model and verify we match HuggingFace output (this one might be harder, so let's see if it's feasbile).

  vllm/tests/models/test_mistral.py

  Lines 12 to 40 in 63e7176

  	@pytest.mark.parametrize("model", MODELS)
  	@pytest.mark.parametrize("dtype", ["bfloat16"])
  	@pytest.mark.parametrize("max_tokens", [128])
  	@pytest.mark.skip(
  	"Two problems: 1. Failing correctness tests. 2. RuntimeError: expected "
  	"scalar type BFloat16 but found Half (only in CI).")
  	def test_models(
  	hf_runner,
  	vllm_runner,
  	example_long_prompts,
  	model: str,
  	dtype: str,
  	max_tokens: int,
  	) -> None:
  	hf_model = hf_runner(model, dtype=dtype)
  	hf_outputs = hf_model.generate_greedy(example_long_prompts, max_tokens)
  	del hf_model
  	vllm_model = vllm_runner(model, dtype=dtype)
  	vllm_outputs = vllm_model.generate_greedy(example_long_prompts, max_tokens)
  	del vllm_model
  	for i in range(len(example_long_prompts)):
  	hf_output_ids, hf_output_str = hf_outputs[i]
  	vllm_output_ids, vllm_output_str = vllm_outputs[i]
  	assert hf_output_str == vllm_output_str, (
  	f"Test{i}:\nHF: {hf_output_str!r}\nvLLM: {vllm_output_str!r}")
  	assert hf_output_ids == vllm_output_ids, (
  	f"Test{i}:\nHF: {hf_output_ids}\nvLLM: {vllm_output_ids}")

[ruthe98]

Thanks for the contribution @ruthe98 . To close this issue I am looking for the following:

• A unit test at the block table level, e.g. at this level
• An e2e test with mistralai/Mistral-7B-v0.1 (sliding window model) that we don't crash when going beyond the sliding window
• An e2e test with the same model and verify we match HuggingFace output (this one might be harder, so let's see if it's feasbile).

  vllm/tests/models/test_mistral.py

  Lines 12 to 40 in 63e7176

  	@pytest.mark.parametrize("model", MODELS)
  	@pytest.mark.parametrize("dtype", ["bfloat16"])
  	@pytest.mark.parametrize("max_tokens", [128])
  	@pytest.mark.skip(
  	"Two problems: 1. Failing correctness tests. 2. RuntimeError: expected "
  	"scalar type BFloat16 but found Half (only in CI).")
  	def test_models(
  	hf_runner,
  	vllm_runner,
  	example_long_prompts,
  	model: str,
  	dtype: str,
  	max_tokens: int,
  	) -> None:
  	hf_model = hf_runner(model, dtype=dtype)
  	hf_outputs = hf_model.generate_greedy(example_long_prompts, max_tokens)
  	del hf_model
  	vllm_model = vllm_runner(model, dtype=dtype)
  	vllm_outputs = vllm_model.generate_greedy(example_long_prompts, max_tokens)
  	del vllm_model
  	for i in range(len(example_long_prompts)):
  	hf_output_ids, hf_output_str = hf_outputs[i]
  	vllm_output_ids, vllm_output_str = vllm_outputs[i]
  	assert hf_output_str == vllm_output_str, (
  	f"Test{i}:\nHF: {hf_output_str!r}\nvLLM: {vllm_output_str!r}")
  	assert hf_output_ids == vllm_output_ids, (
  	f"Test{i}:\nHF: {hf_output_ids}\nvLLM: {vllm_output_ids}")

Ok I'm almost done with this and will send it in soon!

[mmoskal]

@ruthe98 I'm planning to take a closer look on Monday so please share any progress on this, even partial! I have #4405 pending which deals with the kernel side of this

[ruthe98]

Yes I can absolutely push the progress I’ve made! Sorry I’ve been out of town without consistent WiFi but back at it now ¨̮.

…

On Friday, April 26, 2024, Michał Moskal ***@***.***> wrote: @ruthe98 <https://github.com/ruthe98> I'm planning to take a closer look on Monday so please share any progress on this, even partial! I have #4405 <#4405> pending which deals with the kernel side of this — Reply to this email directly, view it on GitHub <#3967 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AHLSRSNI2AAYUPB2SWVEMYLY7LXO7AVCNFSM6AAAAABGAGO2PCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAOBQGI3TANBWGY> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[ruthe98]

I won't have a chance to look at this until later tonight or tomorrow but I've been making some edits. On Sat, Apr 27, 2024 at 2:58 AM Ruth Evans ***@***.***> wrote:

…

Yes I can absolutely push the progress I’ve made! Sorry I’ve been out of town without consistent WiFi but back at it now ¨̮. On Friday, April 26, 2024, Michał Moskal ***@***.***> wrote: > @ruthe98 <https://github.com/ruthe98> I'm planning to take a closer look > on Monday so please share any progress on this, even partial! I have > #4405 <#4405> pending which > deals with the kernel side of this > > — > Reply to this email directly, view it on GitHub > <#3967 (comment)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AHLSRSNI2AAYUPB2SWVEMYLY7LXO7AVCNFSM6AAAAABGAGO2PCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDAOBQGI3TANBWGY> > . > You are receiving this because you were mentioned.Message ID: > ***@***.***> >

[github-actions]

This pull request has been automatically marked as stale because it has not had any activity within 90 days. It will be automatically closed if no further activity occurs within 30 days. Leave a comment if you feel this pull request should remain open. Thank you!

[mergify]

This pull request has merge conflicts that must be resolved before it can be
merged. @ruthe98 please rebase it. https://docs.github.com/en/pull-requests/collaborating-with-pull-requests/working-with-forks/syncing-a-fork