Cleanup readme, llava examples, usage examples and nccl init (#1194)

test/srt/models/test_embedding_models.py

@@ -59,7 +59,7 @@ class TestEmbeddingModels(unittest.TestCase):
                 tolerance = 1e-2
                 assert torch.all(
                     abs(similarities - 1) < tolerance
-                ), f"embeddings not all close"
+                ), "embeddings are not all close"
 
     def test_prefill_logits(self):
         for model, tp_size in MODELS:

test/srt/models/test_generation_models.py

@@ -59,7 +59,7 @@ class TestGenerationModels(unittest.TestCase):
                 tolerance = 3e-2
                 assert torch.all(
                     abs(hf_logprobs - srt_logprobs) < tolerance
-                ), f"prefill logprobs not all close"
+                ), "prefill logprobs are not all close"
 
         print(hf_outputs.output_strs)
         print(srt_outputs.output_strs)