Add Llama 4 to FA3 test (#5509)

Author: hebiao064

.github/workflows/pr-test.yml

@@ -87,6 +87,26 @@ jobs:
           cd test/srt
           python3 run_suite.py --suite per-commit-2-gpu
 
+  unit-test-backend-8-gpu:
+    if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') &&
+        github.event.pull_request.draft == false
+    runs-on: 8-gpu-runner
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+
+      - name: Install dependencies
+        env:
+          FLASHINFER_REPO: ${{ inputs.version == 'nightly' && 'https://flashinfer.ai/whl/nightly/cu124/torch2.5/flashinfer-python' || 'https://flashinfer.ai/whl/cu124/torch2.5/flashinfer-python' }}
+        run: |
+          bash scripts/ci_install_dependency.sh
+
+      - name: Run test
+        timeout-minutes: 30
+        run: |
+          cd test/srt
+          python3 run_suite.py --suite per-commit-8-gpu
+
   performance-test-1-gpu-part-1:
     if: (github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request') &&
         github.event.pull_request.draft == false

python/sglang/test/test_utils.py

@@ -44,7 +44,13 @@
 )
 
 DEFAULT_MODEL_NAME_FOR_TEST = "meta-llama/Llama-3.1-8B-Instruct"
+DEFAULT_MODEL_NAME_FOR_TEST_EAGLE3 = "jamesliu1/sglang-EAGLE3-Llama-3.1-Instruct-8B"
+DEFAULT_MODEL_NAME_FOR_TEST_MLA = "lmsys/sglang-ci-dsv3-test"
+DEFAULT_MODEL_NAME_FOR_TEST_MLA_NEXTN = "lmsys/sglang-ci-dsv3-test-NextN"
 DEFAULT_SMALL_MODEL_NAME_FOR_TEST = "meta-llama/Llama-3.2-1B-Instruct"
+DEFAULT_MODEL_NAME_FOR_TEST_LOCAL_ATTENTION = (
+    "meta-llama/Llama-4-Scout-17B-16E-Instruct"
+)
 DEFAULT_MOE_MODEL_NAME_FOR_TEST = "mistralai/Mixtral-8x7B-Instruct-v0.1"
 DEFAULT_SMALL_MOE_MODEL_NAME_FOR_TEST = "Qwen/Qwen1.5-MoE-A2.7B"
 DEFAULT_SMALL_EMBEDDING_MODEL_NAME_FOR_TEST = "Alibaba-NLP/gte-Qwen2-1.5B-instruct"

test/srt/run_suite.py

@@ -30,7 +30,6 @@ class TestFile:
         TestFile("test_chunked_prefill.py", 336),
         TestFile("test_eagle_infer.py", 500),
         TestFile("test_ebnf_constrained.py"),
-        TestFile("test_fa3.py", 200),
         TestFile("test_fp8_kernel.py", 8),
         TestFile("test_embedding_openai_server.py", 36),
         TestFile("test_hidden_states.py", 55),
@@ -91,6 +90,9 @@ class TestFile:
         TestFile("test_update_weights_from_distributed.py", 100),
         TestFile("test_verl_engine.py", 100),
     ],
+    "per-commit-8-gpu": [
+        TestFile("test_fa3.py", 30),
+    ],
     "nightly": [
         TestFile("test_nightly_gsm8k_eval.py"),
     ],