skip instead of comment

integration-tests/models/test_neox.py

@@ -1,44 +1,46 @@
-# import pytest
+import pytest
-# 
+
-# 
+
-# @pytest.fixture(scope="module")
+@pytest.fixture(scope="module")
-# def neox_handle(launcher):
+def neox_handle(launcher):
-#     with launcher("stabilityai/stablelm-tuned-alpha-3b", num_shard=1, use_flash_attention=False) as handle:
+    with launcher("stabilityai/stablelm-tuned-alpha-3b", num_shard=1, use_flash_attention=False) as handle:
-#         yield handle
+        yield handle
-# 
+
-# 
+
-# @pytest.fixture(scope="module")
+@pytest.fixture(scope="module")
-# async def neox(neox_handle):
+async def neox(neox_handle):
-#     await neox_handle.health(300)
+    await neox_handle.health(300)
-#     return neox_handle.client
+    return neox_handle.client
-# 
+
-# 
+
-# @pytest.mark.asyncio
+@pytest.mark.skip
-# async def test_neox(neox, response_snapshot):
+@pytest.mark.asyncio
-#     response = await neox.generate(
+async def test_neox(neox, response_snapshot):
-#         "<|USER|>What's your mood today?<|ASSISTANT|>",
+    response = await neox.generate(
-#         max_new_tokens=10,
+        "<|USER|>What's your mood today?<|ASSISTANT|>",
-#         decoder_input_details=True,
+        max_new_tokens=10,
-#     )
+        decoder_input_details=True,
-# 
+    )
-#     assert response.details.generated_tokens == 10
+
-#     assert response == response_snapshot
+    assert response.details.generated_tokens == 10
-# 
+    assert response == response_snapshot
-# 
+
-# @pytest.mark.asyncio
+
-# async def test_neox_load(neox, generate_load, response_snapshot):
+@pytest.mark.skip
-#     responses = await generate_load(
+@pytest.mark.asyncio
-#         neox,
+async def test_neox_load(neox, generate_load, response_snapshot):
-#         "<|USER|>What's your mood today?<|ASSISTANT|>",
+    responses = await generate_load(
-#         max_new_tokens=10,
+        neox,
-#         n=4,
+        "<|USER|>What's your mood today?<|ASSISTANT|>",
-#     )
+        max_new_tokens=10,
-# 
+        n=4,
-#     generated_texts = [r.generated_text for r in responses]
+    )
-# 
+
-#     assert len(generated_texts) == 4
+    generated_texts = [r.generated_text for r in responses]
-#     assert generated_texts, all(
+
-#         [text == generated_texts[0] for text in generated_texts]
+    assert len(generated_texts) == 4
-#     )
+    assert generated_texts, all(
-# 
+        [text == generated_texts[0] for text in generated_texts]
-#     assert responses == response_snapshot
+    )
+
+    assert responses == response_snapshot

integration-tests/models/test_neox_sharded.py

@@ -1,40 +1,42 @@
-# import pytest
+import pytest
-# 
+
-# 
+
-# @pytest.fixture(scope="module")
+@pytest.fixture(scope="module")
-# def neox_sharded_handle(launcher):
+def neox_sharded_handle(launcher):
-#     with launcher("OpenAssistant/oasst-sft-1-pythia-12b", num_shard=2, use_flash_attention=False) as handle:
+    with launcher("OpenAssistant/oasst-sft-1-pythia-12b", num_shard=2, use_flash_attention=False) as handle:
-#         yield handle
+        yield handle
-# 
+
-# 
+
-# @pytest.fixture(scope="module")
+@pytest.fixture(scope="module")
-# async def neox_sharded(neox_sharded_handle):
+async def neox_sharded(neox_sharded_handle):
-#     await neox_sharded_handle.health(300)
+    await neox_sharded_handle.health(300)
-#     return neox_sharded_handle.client
+    return neox_sharded_handle.client
-# 
+
-# 
+
-# @pytest.mark.asyncio
+@pytest.mark.skip
-# async def test_neox(neox_sharded, response_snapshot):
+@pytest.mark.asyncio
-#     response = await neox_sharded.generate(
+async def test_neox(neox_sharded, response_snapshot):
-#         "<|prompter|>What is a meme, and what's the history behind this word?<|endoftext|><|assistant|>",
+    response = await neox_sharded.generate(
-#         max_new_tokens=10,
+        "<|prompter|>What is a meme, and what's the history behind this word?<|endoftext|><|assistant|>",
-#         decoder_input_details=True,
+        max_new_tokens=10,
-#     )
+        decoder_input_details=True,
-# 
+    )
-#     assert response.details.generated_tokens == 10
+
-#     assert response == response_snapshot
+    assert response.details.generated_tokens == 10
-# 
+    assert response == response_snapshot
-# 
+
-# @pytest.mark.asyncio
+
-# async def test_neox_load(neox_sharded, generate_load, response_snapshot):
+@pytest.mark.skip
-#     responses = await generate_load(
+@pytest.mark.asyncio
-#         neox_sharded,
+async def test_neox_load(neox_sharded, generate_load, response_snapshot):
-#         "<|prompter|>What is a meme, and what's the history behind this word?<|endoftext|><|assistant|>",
+    responses = await generate_load(
-#         max_new_tokens=10,
+        neox_sharded,
-#         n=4,
+        "<|prompter|>What is a meme, and what's the history behind this word?<|endoftext|><|assistant|>",
-#     )
+        max_new_tokens=10,
-# 
+        n=4,
-#     assert len(responses) == 4
+    )
-#     assert all([r.generated_text == responses[0].generated_text for r in responses])
+
-# 
+    assert len(responses) == 4
-#     assert responses == response_snapshot
+    assert all([r.generated_text == responses[0].generated_text for r in responses])
+
+    assert responses == response_snapshot