Enable merge queue support in CI workflows

.github/workflows/linkcheck.yml

@@ -5,6 +5,7 @@ on:
     branches: [ main, 'release/*' ]
   pull_request:
     branches: [ main, 'release/*' ]
+  merge_group:
 
   # Allows you to run this workflow manually from the Actions tab
   workflow_dispatch:

.github/workflows/quality-check.yaml

@@ -4,6 +4,7 @@ on:
     branches: [ main , 'release/*' ]
   pull_request:
     branches: [ main, 'release/*' ]
+  merge_group:
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}

.github/workflows/ready-label-check.yaml

@@ -9,6 +9,7 @@ on:
       - opened
       - reopened
       - synchronize
+  merge_group:
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}
@@ -18,14 +19,20 @@ jobs:
   ready-label-check:
     runs-on: gcp-k8s-vllm-util
     steps:
+      - name: Auto-succeed for merge queue
+        if: github.event_name == 'merge_group'
+        run: |
+          echo "::info::Merge queue event - automatically passing ready label check"
+          exit 0
+
       - name: Fail if ready label has not been applied to PR
-        if: "!contains(github.event.pull_request.labels.*.name, 'ready')"
+        if: "github.event_name != 'merge_group' && !contains(github.event.pull_request.labels.*.name, 'ready')"
         run: |
           echo "::error::The PR is not labeled as 'ready'"
           exit 1
 
       - name: Succeed if ready label has been applied to PR
-        if: contains(github.event.pull_request.labels.*.name, 'ready')
+        if: "github.event_name != 'merge_group' && contains(github.event.pull_request.labels.*.name, 'ready')"
         run: |
           echo "::info::The PR is labeled as 'ready'"
           exit 0

.github/workflows/test-check-transformers.yaml

@@ -5,6 +5,7 @@ on:
     types: [ labeled, synchronize ]
   push:
     branches: [ main, 'release/*' ]
+  merge_group:
   workflow_dispatch:
     inputs:
       code_coverage:

.github/workflows/test-check.yaml

@@ -4,6 +4,7 @@ on:
     branches: [ main, 'release/*' ]
   push:
     branches: [ main, 'release/*' ]
+  merge_group:
   workflow_dispatch:
     inputs:
       code_coverage:

examples/quantization_w4a4_fp4/qwen3_5_moe.py

@@ -0,0 +1,87 @@
+from transformers import AutoProcessor, Qwen3_5MoeForConditionalGeneration
+from datasets import load_dataset
+from llmcompressor import oneshot
+from llmcompressor.modifiers.quantization import QuantizationModifier
+import torch
+
+MODEL_ID = "/raid/engine/dsikka/models--Qwen--Qwen3.5-397B-A17B/snapshots/7cad2bae11cb49ca79f7d6a0954de2e2756f4e27"
+
+# Load model.
+model = Qwen3_5MoeForConditionalGeneration.from_pretrained(MODEL_ID, dtype="auto")
+processor = AutoProcessor.from_pretrained(MODEL_ID)
+
+
+recipe = QuantizationModifier(
+    targets="Linear",
+    scheme="NVFP4",
+    ignore=[
+        "re:.*lm_head",
+        "re:visual.*",
+        "re:model.visual.*",
+        "re:.*mlp.gate$",
+        "re:.*embed_tokens$",
+        "re:.*shared_expert_gate$",
+        "re:.*mlp\\.shared_expert$",
+        "re:.*linear_attn.*",
+    ],
+)
+
+DATASET_ID = "neuralmagic/calibration"
+NUM_CALIBRATION_SAMPLES = 20
+MAX_SEQUENCE_LENGTH = 8192
+
+ds = load_dataset(DATASET_ID, name="LLM", split=f"train[:{NUM_CALIBRATION_SAMPLES}]")
+
+
+def preprocess_function(example):
+    messgages = []
+    for message in example["messages"]:
+        messgages.append(
+            {
+                "role": message["role"],
+                "content": [{"type": "text", "text": message["content"]}],
+            }
+        )
+
+    return processor.apply_chat_template(
+        messgages,
+        return_tensors="pt",
+        padding=False,
+        truncation=True,
+        max_length=MAX_SEQUENCE_LENGTH,
+        tokenize=True,
+        add_special_tokens=False,
+        return_dict=True,
+        add_generation_prompt=False,
+    )
+
+
+ds = ds.map(preprocess_function, batched=False, remove_columns=ds.column_names)
+
+
+def data_collator(batch):
+    assert len(batch) == 1
+    return {
+        key: (
+            torch.tensor(value)
+            if key != "pixel_values"
+            else torch.tensor(value, dtype=torch.bfloat16).squeeze(0)
+        )
+        for key, value in batch[0].items()
+    }
+
+
+
+# Apply quantization.
+oneshot(model=model, 
+    recipe=recipe, 
+    dataset=ds,
+    data_collator=data_collator,
+    max_seq_length=MAX_SEQUENCE_LENGTH,
+    num_calibration_samples=NUM_CALIBRATION_SAMPLES,
+    moe_calibrate_all_experts=True)
+
+# Save to disk in compressed-tensors format.
+SAVE_DIR = "/raid/engine/dsikka/" + "Qwen3.5-397B-A17B" + "-NVFP4"
+model.save_pretrained(SAVE_DIR)
+processor.save_pretrained(SAVE_DIR)

examples/quantization_w8a8_fp8/qwen3_5_moe.py

@@ -0,0 +1,38 @@
+from transformers import AutoProcessor, Qwen3_5MoeForConditionalGeneration
+
+from llmcompressor import oneshot
+from llmcompressor.modifiers.quantization import QuantizationModifier
+
+MODEL_ID = "/raid/engine/dsikka/models--Qwen--Qwen3.5-397B-A17B/snapshots/7cad2bae11cb49ca79f7d6a0954de2e2756f4e27"
+
+# Load model.
+model = Qwen3_5MoeForConditionalGeneration.from_pretrained(MODEL_ID, dtype="auto")
+processor = AutoProcessor.from_pretrained(MODEL_ID)
+
+# Configure the quantization algorithm and scheme.
+# In this case, we:
+#   * quantize the weights to fp8 with channel-wise quantization
+#   * quantize the activations to fp8 with dynamic token activations
+# NOTE: only datafree quantization is supported for Qwen3-VL-MoE currently
+recipe = QuantizationModifier(
+    targets="Linear",
+    scheme="FP8_DYNAMIC",
+    ignore=[
+        "re:.*lm_head",
+        "re:visual.*",
+        "re:model.visual.*",
+        "re:.*mlp.gate$",
+        "re:.*embed_tokens$",
+        "re:.*shared_expert_gate$",
+        "re:.*mlp\\.shared_expert$",
+        "re:.*linear_attn.*",
+    ],
+)
+
+# Apply quantization.
+oneshot(model=model, recipe=recipe)
+
+# Save to disk in compressed-tensors format.
+SAVE_DIR = "/raid/engine/dsikka/" + "Qwen3.5-397B-A17B" + "-FP8-Dynamic-NoLinearAttn"
+model.save_pretrained(SAVE_DIR)
+processor.save_pretrained(SAVE_DIR)

src/llmcompressor/modeling/__init__.py

@@ -16,6 +16,7 @@
 from .qwen3_moe import CalibrationQwen3MoeSparseMoeBlock  # noqa: F401
 from .qwen3_vl_moe import CalibrateQwen3VLMoeTextSparseMoeBlock  # noqa: F401
 from .qwen3_next_moe import CalibrationQwen3NextSparseMoeBlock  # noqa: F401
+from .qwen3_5_vl_moe import CalibrateQwen3_5MoeTextSparseMoeBlock  # noqa: F401
 # TODO: add granite4, Qwen3Next
 
 from .fuse import *

src/llmcompressor/modeling/qwen3_5_vl_moe.py

@@ -0,0 +1,124 @@
+import torch
+from transformers import Qwen3_5MoeConfig, Qwen3_5MoeTextConfig
+from transformers.models.qwen3_5_moe.modeling_qwen3_5_moe import (
+    Qwen3_5MoeSparseMoeBlock,
+)
+
+from llmcompressor.modeling.moe_context import MoECalibrationModule
+from llmcompressor.utils.dev import skip_weights_initialize
+import torch.nn.functional as F
+
+
+@MoECalibrationModule.register("Qwen3_5MoeSparseMoeBlock")
+class CalibrateQwen3_5MoeTextSparseMoeBlock(MoECalibrationModule):
+    """
+    Calibration version of Qwen3_5MoeSparseMoeBlock that sends all tokens to all
+    experts.
+    """
+
+    is_permanent = True
+
+    def __init__(
+        self,
+        original: "Qwen3_5MoeSparseMoeBlock",
+        config: "Qwen3_5MoeConfig",
+        calibrate_all_experts: bool = True,
+    ):
+        super().__init__()
+        text_config: "Qwen3_5MoeTextConfig" = config.get_text_config()
+
+        self.num_experts = text_config.num_experts
+
+        self.shared_expert = original.shared_expert
+        self.shared_expert_gate = original.shared_expert_gate
+        self.gate = original.gate
+        self.experts = SequentialQwen3VLMoeTextExperts(text_config, original.experts)
+        self.calibrate_all_experts = calibrate_all_experts
+
+    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
+        batch_size, sequence_length, hidden_dim = hidden_states.shape
+        hidden_states_reshaped = hidden_states.view(-1, hidden_dim)
+
+        # router: returns (router_logits, router_scores, router_indices)
+        _, routing_weights, selected_experts = self.gate(hidden_states_reshaped)
+
+        # expert mask: (num_experts, top_k, num_tokens)
+        expert_mask = F.one_hot(selected_experts, num_classes=self.num_experts).permute(
+            2, 1, 0
+        )
+
+        final_hidden_states = torch.zeros(
+            (batch_size * sequence_length, hidden_dim),
+            dtype=hidden_states.dtype,
+            device=hidden_states.device,
+        )
+
+        for expert_idx, expert_layer in enumerate(self.experts):
+            idx, token_idx = torch.where(expert_mask[expert_idx])
+
+            if self.calibrate_all_experts:
+                expert_out = expert_layer(hidden_states_reshaped)[token_idx]
+            else:
+                expert_out = expert_layer(hidden_states_reshaped[token_idx])
+
+            if len(token_idx) > 0:
+                current_hidden_states = (
+                    expert_out * routing_weights[token_idx, idx, None]
+                )
+                final_hidden_states.index_add_(
+                    0,
+                    token_idx,
+                    current_hidden_states.to(hidden_states.dtype),
+                )
+
+        # shared expert
+        shared_expert_output = self.shared_expert(hidden_states_reshaped)
+        shared_expert_output = (
+            F.sigmoid(self.shared_expert_gate(hidden_states_reshaped))
+            * shared_expert_output
+        )
+        final_hidden_states = final_hidden_states + shared_expert_output
+
+        final_hidden_states = final_hidden_states.reshape(
+            batch_size, sequence_length, hidden_dim
+        )
+        return final_hidden_states
+
+    def restore(self, original: torch.nn.Module) -> torch.nn.Module:
+        return original
+
+
+class SequentialQwen3VLMoeTextExperts(torch.nn.ModuleList):
+    def __init__(self, config, original):
+        from transformers.models.qwen3_5_moe.modeling_qwen3_5_moe import (
+            Qwen3_5MoeMLP,
+        )
+        from compressed_tensors.offload import disable_onloading
+
+        self.num_experts = original.gate_up_proj.shape[0]
+        with skip_weights_initialize():
+            super().__init__(
+                [
+                    Qwen3_5MoeMLP(
+                        config, intermediate_size=config.shared_expert_intermediate_size
+                    )
+                    for _ in range(self.num_experts)
+                ]
+            )
+
+        intermediate_size = original.down_proj.shape[-1]
+
+        with disable_onloading():
+            gate_up_data = original.gate_up_proj.data  # [num_experts, 2*inter, hidden]
+            down_data = original.down_proj.data  # [num_experts, hidden, inter]
+
+        for i in range(self.num_experts):
+            gate_up = gate_up_data[i]
+            down = down_data[i]
+
+            gate_proj = gate_up[:intermediate_size, :]
+            up_proj = gate_up[intermediate_size:, :]
+
+            self[i].gate_proj.weight.data = gate_proj.clone().contiguous()
+            self[i].up_proj.weight.data = up_proj.clone().contiguous()
+            self[i].down_proj.weight.data = down.clone().contiguous()

src/llmcompressor/utils/dev.py

@@ -12,7 +12,8 @@
 from loguru import logger
 from safetensors.torch import save_file
 from transformers import AutoModelForCausalLM, PreTrainedModel
-from transformers.modeling_utils import TORCH_INIT_FUNCTIONS
+
+# from transformers.modeling_utils import TORCH_INIT_FUNCTIONS
 from transformers.utils import SAFE_WEIGHTS_INDEX_NAME, WEIGHTS_INDEX_NAME
 
 __all__ = [
@@ -22,6 +23,25 @@
     "dispatch_for_generation",
 ]
 
+from torch import nn
+
+TORCH_INIT_FUNCTIONS = {
+    "uniform_": nn.init.uniform_,
+    "normal_": nn.init.normal_,
+    "trunc_normal_": nn.init.trunc_normal_,
+    "constant_": nn.init.constant_,
+    "xavier_uniform_": nn.init.xavier_uniform_,
+    "xavier_normal_": nn.init.xavier_normal_,
+    "kaiming_uniform_": nn.init.kaiming_uniform_,
+    "kaiming_normal_": nn.init.kaiming_normal_,
+    "uniform": nn.init.uniform,
+    "normal": nn.init.normal,
+    "xavier_uniform": nn.init.xavier_uniform,
+    "xavier_normal": nn.init.xavier_normal,
+    "kaiming_uniform": nn.init.kaiming_uniform,
+    "kaiming_normal": nn.init.kaiming_normal,
+}
+
 
 @contextlib.contextmanager
 def skip_weights_download(model_class: Type[PreTrainedModel] = AutoModelForCausalLM):

src/llmcompressor/utils/pytorch/module.py

@@ -136,7 +136,7 @@ def get_no_split_params(model: PreTrainedModel) -> Union[str, List[str]]:
 
     :return: list of class names that shouldn't be split
     """
-    no_split_modules = model._get_no_split_modules("auto")
+    no_split_modules = model._no_split_modules
     if len(no_split_modules) <= 0:
         return ALL_TARGET