@@ -151,28 +151,20 @@ steps:
151151 exit 0
152152 fi
153153
154- - label : " lora tests for JAX + vLLM models single chip"
154+ - label : " lora e2e tests for JAX + vLLM models single chip"
155155 key : test_10
156156 soft_fail : true
157157 agents :
158158 queue : tpu_v6e_queue
159159 commands :
160160 - |
161- <<<<<<< HEAD
162161 if [[ "$$NIGHTLY" == "1" ]]; then
163162 .buildkite/scripts/run_in_docker.sh \
164- bash -c 'MODEL_IMPL_TYPE=vllm TPU_BACKEND_TYPE=jax python3 -m pytest -s -v -x /workspace/tpu_inference/tests/lora/test_lora.py && \
165- python3 -m pytest -s -v -x /workspace/tpu_inference/tests/lora/test_bgmv.py'
163+ bash -c 'MODEL_IMPL_TYPE=vllm TPU_BACKEND_TYPE=jax python3 -m pytest -s -v -x /workspace/tpu_inference/tests/lora/test_lora.py'
166164 else
167165 echo "Skipping: NIGHTLY environment variable not set"
168166 exit 0
169167 fi
170- =======
171- .buildkite/scripts/run_in_docker.sh \
172- bash -c 'MODEL_IMPL_TYPE=vllm TPU_BACKEND_TYPE=jax python3 -m pytest -s -v -x /workspace/tpu_inference/tests/lora/test_lora.py && \
173- python3 -m pytest -s -v -x /workspace/tpu_inference/tests/lora/test_bgmv.py && \
174- python3 -m pytest -s -v -x /workspace/tpu_inference/tests/lora/test_layers.py'
175- >>>>>>> c17bacea (add multi-chip test case)
176168
177169 - label : " E2E MLPerf tests for JAX + vLLM models on multiple chips"
178170 key : test_11
@@ -210,7 +202,7 @@ steps:
210202 exit 0
211203 fi
212204
213- - label : " lora tests for JAX + vLLM models multi chips"
205+ - label : " lora e2e tests for JAX + vLLM models multi chips"
214206 key : test_13
215207 soft_fail : true
216208 env :
@@ -220,19 +212,13 @@ steps:
220212 queue : tpu_v6e_8_queue
221213 commands :
222214 - |
223- <<<<<<< HEAD
224215 if [[ "$$NIGHTLY" == "1" ]]; then
225216 .buildkite/scripts/run_in_docker.sh \
226217 bash -c 'MODEL_IMPL_TYPE=vllm TPU_BACKEND_TYPE=jax python3 -m pytest -s -v -x /workspace/tpu_inference/tests/lora/test_lora.py'
227218 else
228219 echo "Skipping: NIGHTLY environment variable not set"
229220 exit 0
230221 fi
231- =======
232- .buildkite/scripts/run_in_docker.sh \
233- bash -c 'MODEL_IMPL_TYPE=vllm TPU_BACKEND_TYPE=jax python3 -m pytest -s -v -x /workspace/tpu_inference/tests/lora/test_lora.py && \
234- python3 -m pytest -s -v -x /workspace/tpu_inference/tests/lora/test_layers.py'
235- >>>>>>> c17bacea (add multi-chip test case)
236222
237223 - label : " E2E data parallelism test"
238224 key : test_14
@@ -246,6 +232,29 @@ steps:
246232 .buildkite/scripts/run_in_docker.sh \
247233 bash -c 'python3 -m pytest -s -v -x /workspace/tpu_inference/tests/e2e/test_data_parallel.py'
248234
235+ - label : " lora unit tests on single chip"
236+ key : test_15
237+ soft_fail : true
238+ agents :
239+ queue : tpu_v6e_queue
240+ commands :
241+ - |
242+ .buildkite/scripts/run_in_docker.sh \
243+ bash -c ' python3 -m pytest -s -v -x /workspace/tpu_inference/tests/lora/test_bgmv.py && \
244+ python3 -m pytest -s -v -x /workspace/tpu_inference/tests/lora/test_layers.py'
245+
246+ - label : " lora unit tests on multi chips"
247+ key : test_16
248+ soft_fail : true
249+ env :
250+ USE_V6E8_QUEUE : " True"
251+ VLLM_LOG_LEVEL : " INFO"
252+ agents :
253+ queue : tpu_v6e_8_queue
254+ commands :
255+ - |
256+ .buildkite/scripts/run_in_docker.sh \
257+ bash -c 'python3 -m pytest -s -v -x /workspace/tpu_inference/tests/lora/test_layers.py'
249258 # -----------------------------------------------------------------
250259 # NOTIFICATION STEP
251260 # -----------------------------------------------------------------
@@ -266,9 +275,11 @@ steps:
266275 - test_12
267276 - test_13
268277 - test_14
278+ - test_15
279+ - test_16
269280 agents :
270281 queue : cpu
271282 commands :
272283 - |
273284 .buildkite/scripts/check_results.sh \
274- "TPU JAX Tests Failed" test_0 test_1 test_2 test_3 test_4 test_5 test_6 test_7 test_8 test_9 test_10 test_11 test_12 test_13
285+ "TPU JAX Tests Failed" test_0 test_1 test_2 test_3 test_4 test_5 test_6 test_7 test_8 test_9 test_10 test_11 test_12 test_13 test_14 test_15 test_16
0 commit comments