test(ascend): validate documented Qwen3.6 models

Author: hooray03

.github/scripts/ascend/download_models.sh

@@ -21,19 +21,24 @@ export FL_MODEL_BASE_PATH="${FL_MODEL_BASE_PATH:-/data/models}"
 export HF_ENDPOINT="${HF_ENDPOINT:-https://hf-mirror.com}"
 
 QWEN_ROOT="${FL_MODEL_BASE_PATH}/Qwen"
-MODEL_ID="Qwen/Qwen3-0.6B"
-MODEL_DIR="${QWEN_ROOT}/Qwen3-0.6B"
+MODEL_IDS=(
+    "Qwen/Qwen3-0.6B"
+    "Qwen/Qwen3.6-27B"
+    "Qwen/Qwen3.6-35B-A3B"
+)
 
 mkdir -p "${QWEN_ROOT}"
 
-if [[ -f "${MODEL_DIR}/config.json" ]]; then
-    echo "Model already available: ${MODEL_DIR}"
-    exit 0
-fi
+for MODEL_ID in "${MODEL_IDS[@]}"; do
+    MODEL_DIR="${QWEN_ROOT}/${MODEL_ID#Qwen/}"
+    if [[ -f "${MODEL_DIR}/config.json" ]]; then
+        echo "Model already available: ${MODEL_DIR}"
+        continue
+    fi
 
-export MODEL_ID MODEL_DIR
-echo "Downloading ${MODEL_ID} from ${HF_ENDPOINT} to ${MODEL_DIR}"
-python - <<'PY'
+    export MODEL_ID MODEL_DIR
+    echo "Downloading ${MODEL_ID} from ${HF_ENDPOINT} to ${MODEL_DIR}"
+    python - <<'PY'
 import os
 
 from huggingface_hub import snapshot_download
@@ -45,5 +50,6 @@ snapshot_download(
 )
 PY
 
-test -f "${MODEL_DIR}/config.json"
-echo "Model ready: ${MODEL_DIR}"
+    test -f "${MODEL_DIR}/config.json"
+    echo "Model ready: ${MODEL_DIR}"
+done

tests/models/qwen3_6/27b_tp2_eager_ascend.yaml

@@ -0,0 +1,46 @@
+# Copyright 2026 FlagOS Contributors
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# Qwen3.6-27B Ascend 910C validation from the platform verification guide.
+
+llm:
+  model: "/data/models/Qwen/Qwen3.6-27B"
+  tensor_parallel_size: 2
+  pipeline_parallel_size: 1
+  max_model_len: 4096
+  gpu_memory_utilization: 0.8
+  enforce_eager: true
+  trust_remote_code: true
+  enable_chunked_prefill: false
+  async_scheduling: false
+  enable_prefix_caching: false
+
+generate:
+  prompts:
+    - "Introduce large language models briefly."
+  sampling:
+    max_tokens: 100
+    temperature: 0.0
+
+serve:
+  served_model_name: "qwen"
+  startup_retries: 180
+  endpoints: ["chat"]
+  stream: true
+  max_tokens: 1024
+  chat_messages:
+    - role: "user"
+      content: "Introduce large language models briefly."
+  sampling:
+    temperature: 0.0

tests/models/qwen3_6/35b_a3b_tp2_eager_ascend.yaml

@@ -0,0 +1,46 @@
+# Copyright 2026 FlagOS Contributors
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+# Qwen3.6-35B-A3B Ascend 910C validation from the platform verification guide.
+
+llm:
+  model: "/data/models/Qwen/Qwen3.6-35B-A3B"
+  tensor_parallel_size: 2
+  pipeline_parallel_size: 1
+  max_model_len: 4096
+  gpu_memory_utilization: 0.8
+  enforce_eager: true
+  trust_remote_code: true
+  enable_chunked_prefill: false
+  async_scheduling: false
+  enable_prefix_caching: false
+
+generate:
+  prompts:
+    - "Introduce large language models briefly."
+  sampling:
+    max_tokens: 100
+    temperature: 0.0
+
+serve:
+  served_model_name: "qwen"
+  startup_retries: 180
+  endpoints: ["chat"]
+  stream: true
+  max_tokens: 1024
+  chat_messages:
+    - role: "user"
+      content: "Introduce large language models briefly."
+  sampling:
+    temperature: 0.0

tests/platforms/ascend.yaml

@@ -63,9 +63,9 @@ env_defaults: {}
     e2e:
       # Add or remove test cases per model
       inference:
-        qwen3: ["06b_tp2"]
+        qwen3_6: ["27b_tp2_eager_ascend", "35b_a3b_tp2_eager_ascend"]
       serving:
-        qwen3: ["06b_tp2"]
+        qwen3_6: ["27b_tp2_eager_ascend", "35b_a3b_tp2_eager_ascend"]
     functional:
       # Include patterns: "*" for all, or list specific paths
       include: "*"