[TPU][Test] Rollback PR-21550. (vllm-project#21619)

Signed-off-by: Qiliang Cui <[email protected]>
Signed-off-by: x22x22 <[email protected]>

Author: QiliangCui

tests/v1/tpu/test_basic.py

@@ -59,7 +59,7 @@ def test_basic(
                 # actually test chunked prompt
                 max_num_batched_tokens=1024,
                 max_model_len=8192,
-                gpu_memory_utilization=0.95,
+                gpu_memory_utilization=0.7,
                 max_num_seqs=max_num_seqs,
                 tensor_parallel_size=tensor_parallel_size) as vllm_model:
             vllm_outputs = vllm_model.generate_greedy(example_prompts,