[Model][1/N] Delete deepseek v2/v3 modeling codes.#3189
Merged
wangxiyuan merged 6 commits intovllm-project:mainfrom Oct 20, 2025
Merged
[Model][1/N] Delete deepseek v2/v3 modeling codes.#3189wangxiyuan merged 6 commits intovllm-project:mainfrom
wangxiyuan merged 6 commits intovllm-project:mainfrom
Conversation
Contributor
|
👋 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. |
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Code Review
This pull request deletes the file vllm_ascend/models/deepseek_v2.py, which contains the non-torchair implementations for DeepSeek V2 and V3 models. While the goal seems to be cleanup, this change is incomplete and introduces breaking changes. Specifically, vllm_ascend/models/__init__.py still attempts to register models from the deleted file, which will cause an ImportError. Additionally, vllm_ascend/models/deepseek_mtp.py relies on a monkey patch from the deleted file, and its behavior will change, potentially leading to correctness or performance issues. The test file tests/ut/models/test_deepseek_v2.py also has a direct dependency on the deleted file and will fail. This pull request should be updated to resolve these dangling dependencies.
whx-sjtu
force-pushed
the
rm_ds_model
branch
from
5c376dd to
1a341b7
Compare
whx-sjtu
force-pushed
the
rm_ds_model
branch
3 times, most recently
from
1730fb5 to
e25f3ed
Compare
whx-sjtu
force-pushed
the
rm_ds_model
branch
4 times, most recently
from
1d78a51 to
0b1e706
Compare
whx-sjtu
force-pushed
the
rm_ds_model
branch
2 times, most recently
from
71be184 to
380d3ef
Compare
Contributor
|
This pull request has conflicts, please resolve those before we can evaluate the pull request. |
Collaborator
|
Why is it possible to delete these two model files now? |
Contributor
|
This pull request has conflicts, please resolve those before we can evaluate the pull request. |
whx-sjtu
changed the title
Collaborator
Author
What do you think is the reason why it cannot be deleted? If you refer to the CUDA hard-coding issue of DeepSeek 3.2, I will circumvent it through a patch later. @LCAIZJ |
Contributor
|
This pull request has conflicts, please resolve those before we can evaluate the pull request. |
Signed-off-by: whx-sjtu <2952154980@qq.com>
Signed-off-by: whx-sjtu <2952154980@qq.com>
Signed-off-by: whx-sjtu <2952154980@qq.com>
Signed-off-by: whx-sjtu <2952154980@qq.com>
whx-sjtu
changed the title
whx-sjtu
changed the title
Collaborator
|
Nice work |
wangxiyuan
approved these changes
Oct 20, 2025
ZYang6263
pushed a commit
to rjg-lyh/vllm-ascend
that referenced
this pull request
Oct 23, 2025
This PR deletes model codes of deepseek_v2 and deepseek_v3 to reuse the model file from vLLM. vLLM Ascend now uses custom ops register way instead of model file hard-coding. - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 --------- Signed-off-by: whx-sjtu <2952154980@qq.com>
wangxiyuan
pushed a commit
that referenced
this pull request
Oct 30, 2025
This is the follow-up PR to PR #3189, which continues to refactor sfa into mla and finally remove deepseek_v3_2.py. This is the last PR of deepseek modeling refactoring. After this, all deepseek-related model codes are removed from vllm_ascend. FurtherMore, after this PR deepseek v3.2 can run chunk-prefill with correct accuracy. - vLLM version: v0.11.0rc3 - vLLM main: vllm-project/vllm@83f478b --------- Signed-off-by: whx-sjtu <2952154980@qq.com>
luolun
pushed a commit
to luolun/vllm-ascend
that referenced
this pull request
Nov 19, 2025
This PR deletes model codes of deepseek_v2 and deepseek_v3 to reuse the model file from vLLM. vLLM Ascend now uses custom ops register way instead of model file hard-coding. - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 --------- Signed-off-by: whx-sjtu <2952154980@qq.com> Signed-off-by: luolun <luolun1995@cmbchina.com>
luolun
pushed a commit
to luolun/vllm-ascend
that referenced
this pull request
Nov 19, 2025
…project#3769) This is the follow-up PR to PR vllm-project#3189, which continues to refactor sfa into mla and finally remove deepseek_v3_2.py. This is the last PR of deepseek modeling refactoring. After this, all deepseek-related model codes are removed from vllm_ascend. FurtherMore, after this PR deepseek v3.2 can run chunk-prefill with correct accuracy. - vLLM version: v0.11.0rc3 - vLLM main: vllm-project/vllm@83f478b --------- Signed-off-by: whx-sjtu <2952154980@qq.com> Signed-off-by: luolun <luolun1995@cmbchina.com>
hwhaokun
pushed a commit
to hwhaokun/vllm-ascend
that referenced
this pull request
Nov 19, 2025
This PR deletes model codes of deepseek_v2 and deepseek_v3 to reuse the model file from vLLM. vLLM Ascend now uses custom ops register way instead of model file hard-coding. - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 --------- Signed-off-by: whx-sjtu <2952154980@qq.com> Signed-off-by: hwhaokun <haokun0405@163.com>
hwhaokun
pushed a commit
to hwhaokun/vllm-ascend
that referenced
this pull request
Nov 19, 2025
…project#3769) This is the follow-up PR to PR vllm-project#3189, which continues to refactor sfa into mla and finally remove deepseek_v3_2.py. This is the last PR of deepseek modeling refactoring. After this, all deepseek-related model codes are removed from vllm_ascend. FurtherMore, after this PR deepseek v3.2 can run chunk-prefill with correct accuracy. - vLLM version: v0.11.0rc3 - vLLM main: vllm-project/vllm@83f478b --------- Signed-off-by: whx-sjtu <2952154980@qq.com> Signed-off-by: hwhaokun <haokun0405@163.com>
NSDie
pushed a commit
to NSDie/vllm-ascend
that referenced
this pull request
Nov 24, 2025
This PR deletes model codes of deepseek_v2 and deepseek_v3 to reuse the model file from vLLM. vLLM Ascend now uses custom ops register way instead of model file hard-coding. - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 --------- Signed-off-by: whx-sjtu <2952154980@qq.com> Signed-off-by: nsdie <yeyifan@huawei.com>
NSDie
pushed a commit
to NSDie/vllm-ascend
that referenced
this pull request
Nov 24, 2025
…project#3769) This is the follow-up PR to PR vllm-project#3189, which continues to refactor sfa into mla and finally remove deepseek_v3_2.py. This is the last PR of deepseek modeling refactoring. After this, all deepseek-related model codes are removed from vllm_ascend. FurtherMore, after this PR deepseek v3.2 can run chunk-prefill with correct accuracy. - vLLM version: v0.11.0rc3 - vLLM main: vllm-project/vllm@83f478b --------- Signed-off-by: whx-sjtu <2952154980@qq.com> Signed-off-by: nsdie <yeyifan@huawei.com>
Clorist33
pushed a commit
to Clorist33/vllm-ascend
that referenced
this pull request
Dec 9, 2025
This PR deletes model codes of deepseek_v2 and deepseek_v3 to reuse the model file from vLLM. vLLM Ascend now uses custom ops register way instead of model file hard-coding. - vLLM version: v0.11.0rc3 - vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.0 --------- Signed-off-by: whx-sjtu <2952154980@qq.com>
Clorist33
pushed a commit
to Clorist33/vllm-ascend
that referenced
this pull request
Dec 10, 2025
…project#3769) This is the follow-up PR to PR vllm-project#3189, which continues to refactor sfa into mla and finally remove deepseek_v3_2.py. This is the last PR of deepseek modeling refactoring. After this, all deepseek-related model codes are removed from vllm_ascend. FurtherMore, after this PR deepseek v3.2 can run chunk-prefill with correct accuracy. - vLLM version: v0.11.0rc3 - vLLM main: vllm-project/vllm@83f478b --------- Signed-off-by: whx-sjtu <2952154980@qq.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR deletes model codes of deepseek_v2 and deepseek_v3 to reuse the model file from vLLM.
vLLM Ascend now uses custom ops register way instead of model file hard-coding.