Fix target matching for fused layers with compressed-tensors #12617

eldarkurtic · 2025-01-31T16:35:01Z

Without this PR

Quantizing models with llm-compressor and a recipe that explicitly lists names of layers produces a model that is not loadable by vLLM (i.e. vllm serve <model> fails with raise ValueError(f"Unable to find matching target for {module} in the ...).

Example recipe:

recipe = """
quantization_stage:
  run_type: oneshot
  quantization_modifiers:
    GPTQModifier:
      ignore: ["lm_head"]
      config_groups:
        group_0:
          weights:
            num_bits: 4
            type: "int"
            symmetric: true
            strategy: "group"
            group_size: 128
          targets: [
            "model.layers.0.mlp.down_proj",
            "model.layers.2.mlp.down_proj",
            "model.layers.3.mlp.down_proj",
            "model.layers.4.mlp.down_proj",
            "model.layers.5.mlp.down_proj",
            "model.layers.6.mlp.down_proj",
            "model.layers.7.mlp.down_proj",
            "model.layers.8.mlp.down_proj",
            "model.layers.9.mlp.down_proj",
            "model.layers.10.mlp.down_proj",
            "model.layers.11.mlp.down_proj",
            "model.layers.12.mlp.down_proj",
            "model.layers.13.mlp.down_proj",
            "model.layers.14.mlp.down_proj",
            "model.layers.15.mlp.down_proj",
            "model.layers.16.mlp.down_proj",
            "model.layers.17.mlp.down_proj",
            "model.layers.19.mlp.down_proj",
            "model.layers.21.mlp.down_proj",
            "model.layers.22.mlp.down_proj",
            .
            .
            .
          ]
"""

To reproduce the vLLM error:

vllm serve nm-testing/eldar-test

With this PR

Models are loaded correctly without any errors.

github-actions · 2025-01-31T16:35:14Z

👋 Hi! Thank you for contributing to the vLLM project.
Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can do one of these:

Add ready label to the PR
Enable auto-merge.

🚀

vllm/model_executor/layers/quantization/compressed_tensors/utils.py

mgoin

LGTM!

…oject#12617) Without this PR --------------- Quantizing models with llm-compressor and a recipe that explicitly lists names of layers produces a model that is not loadable by vLLM (i.e. `vllm serve <model>` fails with `raise ValueError(f"Unable to find matching target for {module} in the ...`). Example recipe: ``` recipe = """ quantization_stage: run_type: oneshot quantization_modifiers: GPTQModifier: ignore: ["lm_head"] config_groups: group_0: weights: num_bits: 4 type: "int" symmetric: true strategy: "group" group_size: 128 targets: [ "model.layers.0.mlp.down_proj", "model.layers.2.mlp.down_proj", "model.layers.3.mlp.down_proj", "model.layers.4.mlp.down_proj", "model.layers.5.mlp.down_proj", "model.layers.6.mlp.down_proj", "model.layers.7.mlp.down_proj", "model.layers.8.mlp.down_proj", "model.layers.9.mlp.down_proj", "model.layers.10.mlp.down_proj", "model.layers.11.mlp.down_proj", "model.layers.12.mlp.down_proj", "model.layers.13.mlp.down_proj", "model.layers.14.mlp.down_proj", "model.layers.15.mlp.down_proj", "model.layers.16.mlp.down_proj", "model.layers.17.mlp.down_proj", "model.layers.19.mlp.down_proj", "model.layers.21.mlp.down_proj", "model.layers.22.mlp.down_proj", . . . ] """ ``` To reproduce the vLLM error: ```bash vllm serve nm-testing/eldar-test ``` With this PR ------------ Models are loaded correctly without any errors. Signed-off-by: Isotr0py <2037008807@qq.com>

kylesayrs · 2025-02-02T16:24:31Z

This seems to be ignoring the existing logic in _handle_fused_layers. The _handle_fused_layers function is insufficient, as it only maps 1 to 1, so it should be removed

…oject#12617) Without this PR --------------- Quantizing models with llm-compressor and a recipe that explicitly lists names of layers produces a model that is not loadable by vLLM (i.e. `vllm serve <model>` fails with `raise ValueError(f"Unable to find matching target for {module} in the ...`). Example recipe: ``` recipe = """ quantization_stage: run_type: oneshot quantization_modifiers: GPTQModifier: ignore: ["lm_head"] config_groups: group_0: weights: num_bits: 4 type: "int" symmetric: true strategy: "group" group_size: 128 targets: [ "model.layers.0.mlp.down_proj", "model.layers.2.mlp.down_proj", "model.layers.3.mlp.down_proj", "model.layers.4.mlp.down_proj", "model.layers.5.mlp.down_proj", "model.layers.6.mlp.down_proj", "model.layers.7.mlp.down_proj", "model.layers.8.mlp.down_proj", "model.layers.9.mlp.down_proj", "model.layers.10.mlp.down_proj", "model.layers.11.mlp.down_proj", "model.layers.12.mlp.down_proj", "model.layers.13.mlp.down_proj", "model.layers.14.mlp.down_proj", "model.layers.15.mlp.down_proj", "model.layers.16.mlp.down_proj", "model.layers.17.mlp.down_proj", "model.layers.19.mlp.down_proj", "model.layers.21.mlp.down_proj", "model.layers.22.mlp.down_proj", . . . ] """ ``` To reproduce the vLLM error: ```bash vllm serve nm-testing/eldar-test ``` With this PR ------------ Models are loaded correctly without any errors.

…oject#12617) Without this PR --------------- Quantizing models with llm-compressor and a recipe that explicitly lists names of layers produces a model that is not loadable by vLLM (i.e. `vllm serve <model>` fails with `raise ValueError(f"Unable to find matching target for {module} in the ...`). Example recipe: ``` recipe = """ quantization_stage: run_type: oneshot quantization_modifiers: GPTQModifier: ignore: ["lm_head"] config_groups: group_0: weights: num_bits: 4 type: "int" symmetric: true strategy: "group" group_size: 128 targets: [ "model.layers.0.mlp.down_proj", "model.layers.2.mlp.down_proj", "model.layers.3.mlp.down_proj", "model.layers.4.mlp.down_proj", "model.layers.5.mlp.down_proj", "model.layers.6.mlp.down_proj", "model.layers.7.mlp.down_proj", "model.layers.8.mlp.down_proj", "model.layers.9.mlp.down_proj", "model.layers.10.mlp.down_proj", "model.layers.11.mlp.down_proj", "model.layers.12.mlp.down_proj", "model.layers.13.mlp.down_proj", "model.layers.14.mlp.down_proj", "model.layers.15.mlp.down_proj", "model.layers.16.mlp.down_proj", "model.layers.17.mlp.down_proj", "model.layers.19.mlp.down_proj", "model.layers.21.mlp.down_proj", "model.layers.22.mlp.down_proj", . . . ] """ ``` To reproduce the vLLM error: ```bash vllm serve nm-testing/eldar-test ``` With this PR ------------ Models are loaded correctly without any errors. Signed-off-by: Srikanth Srinivas <srikanth@astrum.ai>

…oject#12617) Without this PR --------------- Quantizing models with llm-compressor and a recipe that explicitly lists names of layers produces a model that is not loadable by vLLM (i.e. `vllm serve <model>` fails with `raise ValueError(f"Unable to find matching target for {module} in the ...`). Example recipe: ``` recipe = """ quantization_stage: run_type: oneshot quantization_modifiers: GPTQModifier: ignore: ["lm_head"] config_groups: group_0: weights: num_bits: 4 type: "int" symmetric: true strategy: "group" group_size: 128 targets: [ "model.layers.0.mlp.down_proj", "model.layers.2.mlp.down_proj", "model.layers.3.mlp.down_proj", "model.layers.4.mlp.down_proj", "model.layers.5.mlp.down_proj", "model.layers.6.mlp.down_proj", "model.layers.7.mlp.down_proj", "model.layers.8.mlp.down_proj", "model.layers.9.mlp.down_proj", "model.layers.10.mlp.down_proj", "model.layers.11.mlp.down_proj", "model.layers.12.mlp.down_proj", "model.layers.13.mlp.down_proj", "model.layers.14.mlp.down_proj", "model.layers.15.mlp.down_proj", "model.layers.16.mlp.down_proj", "model.layers.17.mlp.down_proj", "model.layers.19.mlp.down_proj", "model.layers.21.mlp.down_proj", "model.layers.22.mlp.down_proj", . . . ] """ ``` To reproduce the vLLM error: ```bash vllm serve nm-testing/eldar-test ``` With this PR ------------ Models are loaded correctly without any errors.

…oject#12617) Without this PR --------------- Quantizing models with llm-compressor and a recipe that explicitly lists names of layers produces a model that is not loadable by vLLM (i.e. `vllm serve <model>` fails with `raise ValueError(f"Unable to find matching target for {module} in the ...`). Example recipe: ``` recipe = """ quantization_stage: run_type: oneshot quantization_modifiers: GPTQModifier: ignore: ["lm_head"] config_groups: group_0: weights: num_bits: 4 type: "int" symmetric: true strategy: "group" group_size: 128 targets: [ "model.layers.0.mlp.down_proj", "model.layers.2.mlp.down_proj", "model.layers.3.mlp.down_proj", "model.layers.4.mlp.down_proj", "model.layers.5.mlp.down_proj", "model.layers.6.mlp.down_proj", "model.layers.7.mlp.down_proj", "model.layers.8.mlp.down_proj", "model.layers.9.mlp.down_proj", "model.layers.10.mlp.down_proj", "model.layers.11.mlp.down_proj", "model.layers.12.mlp.down_proj", "model.layers.13.mlp.down_proj", "model.layers.14.mlp.down_proj", "model.layers.15.mlp.down_proj", "model.layers.16.mlp.down_proj", "model.layers.17.mlp.down_proj", "model.layers.19.mlp.down_proj", "model.layers.21.mlp.down_proj", "model.layers.22.mlp.down_proj", . . . ] """ ``` To reproduce the vLLM error: ```bash vllm serve nm-testing/eldar-test ``` With this PR ------------ Models are loaded correctly without any errors. Signed-off-by: Felix Marty <felmarty@amd.com>

…oject#12617) Without this PR --------------- Quantizing models with llm-compressor and a recipe that explicitly lists names of layers produces a model that is not loadable by vLLM (i.e. `vllm serve <model>` fails with `raise ValueError(f"Unable to find matching target for {module} in the ...`). Example recipe: ``` recipe = """ quantization_stage: run_type: oneshot quantization_modifiers: GPTQModifier: ignore: ["lm_head"] config_groups: group_0: weights: num_bits: 4 type: "int" symmetric: true strategy: "group" group_size: 128 targets: [ "model.layers.0.mlp.down_proj", "model.layers.2.mlp.down_proj", "model.layers.3.mlp.down_proj", "model.layers.4.mlp.down_proj", "model.layers.5.mlp.down_proj", "model.layers.6.mlp.down_proj", "model.layers.7.mlp.down_proj", "model.layers.8.mlp.down_proj", "model.layers.9.mlp.down_proj", "model.layers.10.mlp.down_proj", "model.layers.11.mlp.down_proj", "model.layers.12.mlp.down_proj", "model.layers.13.mlp.down_proj", "model.layers.14.mlp.down_proj", "model.layers.15.mlp.down_proj", "model.layers.16.mlp.down_proj", "model.layers.17.mlp.down_proj", "model.layers.19.mlp.down_proj", "model.layers.21.mlp.down_proj", "model.layers.22.mlp.down_proj", . . . ] """ ``` To reproduce the vLLM error: ```bash vllm serve nm-testing/eldar-test ``` With this PR ------------ Models are loaded correctly without any errors.

Properly match fused layers

589a540

eldarkurtic requested review from mgoin, robertgshaw2-redhat and tlrmchlsmth as code owners January 31, 2025 16:35

mgoin changed the title ~~Fix matching for fused layers~~ Fix target matching for fused layers with compressed-tensors Jan 31, 2025

mgoin added quantization ready ONLY add when PR is ready to merge/full CI is needed labels Jan 31, 2025

eldarkurtic added 2 commits January 31, 2025 16:49

fix line length

e2f9563

fix style

777baae

mgoin reviewed Jan 31, 2025

View reviewed changes

vllm/model_executor/layers/quantization/compressed_tensors/utils.py Outdated Show resolved Hide resolved

make sure all fused layers have the same quant

cf2511c

eldarkurtic requested a review from mgoin January 31, 2025 21:44

mgoin approved these changes Jan 31, 2025

View reviewed changes

mgoin enabled auto-merge (squash) February 1, 2025 00:38

mgoin merged commit 1867c25 into vllm-project:main Feb 1, 2025
49 checks passed

kylesayrs mentioned this pull request Feb 2, 2025

[Model][Quant] Fix GLM, Fix fused module mappings for quantization #12634

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix target matching for fused layers with compressed-tensors #12617

Fix target matching for fused layers with compressed-tensors #12617

eldarkurtic commented Jan 31, 2025 •

edited by github-actions bot

Loading

github-actions bot commented Jan 31, 2025

mgoin left a comment

kylesayrs commented Feb 2, 2025

Fix target matching for fused layers with compressed-tensors #12617

Fix target matching for fused layers with compressed-tensors #12617

Conversation

eldarkurtic commented Jan 31, 2025 • edited by github-actions bot Loading

Without this PR

With this PR

github-actions bot commented Jan 31, 2025

mgoin left a comment

Choose a reason for hiding this comment

kylesayrs commented Feb 2, 2025

eldarkurtic commented Jan 31, 2025 •

edited by github-actions bot

Loading