Skip to content

Commit a718486

Browse files
authored
fix: Fix DeepSeek R1 CI (#6129)
Signed-off-by: Yi Zhang <[email protected]>
1 parent 9b45499 commit a718486

File tree

2 files changed

+2
-4
lines changed

2 files changed

+2
-4
lines changed

tests/integration/defs/accuracy/test_llm_api_pytorch.py

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -1352,7 +1352,7 @@ def test_nvfp4_multi_gpus(self, tp_size, pp_size, ep_size, mtp_nextn, fp8kv,
13521352
attention_dp, cuda_graph, overlap_scheduler,
13531353
max_batch_size, moe_backend):
13541354

1355-
kv_cache_config = KvCacheConfig(free_gpu_memory_fraction=0.85)
1355+
kv_cache_config = KvCacheConfig(free_gpu_memory_fraction=0.80)
13561356
pytorch_config = dict(
13571357
disable_overlap_scheduler=not overlap_scheduler,
13581358
cuda_graph_config=CudaGraphConfig() if cuda_graph else None,
@@ -1374,7 +1374,7 @@ def test_nvfp4_multi_gpus(self, tp_size, pp_size, ep_size, mtp_nextn, fp8kv,
13741374
enable_attention_dp=attention_dp,
13751375
speculative_config=mtp_config) as llm:
13761376

1377-
assert llm.args.moe_backend == moe_backend
1377+
assert llm.args.moe_config.backend == moe_backend
13781378
assert llm.args.quant_config.quant_algo == QuantAlgo.NVFP4
13791379

13801380
task = MMLU(self.MODEL_NAME)

tests/integration/test_lists/waives.txt

Lines changed: 0 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -439,5 +439,3 @@ examples/test_multimodal.py::test_llm_multimodal_general[fuyu-8b-pp:1-tp:1-float
439439
test_e2e.py::test_ptp_quickstart SKIP (https://nvbugs/5387762)
440440
triton_server/test_triton_llm.py::test_llava_onevision[test_basic-False-1---False-True-False-0-128-disableDecoupleMode-inflight_fused_batching-disableTrtOverlap-0.2-max_utilization---1-1-1-False-tensorrt_llm_bls] SKIP (https://nvbugs/5396437)
441441
triton_server/test_triton_llm.py::test_llava_onevision[test_video-False-1---False-True-False-0-128-disableDecoupleMode-inflight_fused_batching-disableTrtOverlap-0.2-guaranteed_no_evict---1-1-1-False-tensorrt_llm_bls] SKIP (https://nvbugs/5396437)
442-
accuracy/test_llm_api_pytorch.py::TestDeepSeekR1::test_nvfp4_multi_gpus[latency] SKIP (https://nvbugs/5397036)
443-
accuracy/test_llm_api_pytorch.py::TestDeepSeekR1::test_nvfp4_multi_gpus[throughput_tp8] SKIP (https://nvbugs/5397036)

0 commit comments

Comments
 (0)