mirror of
https://github.com/huggingface/text-generation-inference.git
synced 2025-04-19 22:02:06 +00:00
93a7042d7e
* feat: support phi3.5 moe model loading * fix: prefer llama base model and improve rotary logic * feat: return reasonable generation and add integration test * fix: run lint and update docs * fix: rerun lint for openapi docs * fix: prefer do_sample false unless temp is set by user, and update chat tests * fix: small typo adjustments * fix: consolidate long rope paths * fix: revert greedy by default and test changes * Vendor configuration so that we don't have to `trust_remote_code` * Use SparseMoELayer * Add support for dense MoE * Some type annotations * Add the usual model tests * Ruff. --------- Co-authored-by: Daniël de Kok <me@danieldk.eu> Co-authored-by: Nicolas Patry <patry.nicolas@protonmail.com>
76 lines
2.1 KiB
Python
76 lines
2.1 KiB
Python
import pytest
|
|
|
|
|
|
@pytest.fixture(scope="module")
|
|
def flash_phi35_moe_handle(launcher):
|
|
with launcher(
|
|
"microsoft/Phi-3.5-MoE-instruct",
|
|
num_shard=4,
|
|
) as handle:
|
|
yield handle
|
|
|
|
|
|
@pytest.fixture(scope="module")
|
|
async def flash_phi35_moe(flash_phi35_moe_handle):
|
|
await flash_phi35_moe_handle.health(300)
|
|
return flash_phi35_moe_handle.client
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_flash_phi35_moe(flash_phi35_moe, response_snapshot):
|
|
response = await flash_phi35_moe.generate(
|
|
"What is gradient descent?\n\n", max_new_tokens=10, decoder_input_details=True
|
|
)
|
|
|
|
assert response.details.generated_tokens == 10
|
|
assert (
|
|
response.generated_text
|
|
== "Gradient descent is a first-order optimization algorithm"
|
|
)
|
|
assert response == response_snapshot
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_flash_phi35_moe_all_params(flash_phi35_moe, response_snapshot):
|
|
response = await flash_phi35_moe.generate(
|
|
"What is gradient descent?\n\n",
|
|
max_new_tokens=10,
|
|
repetition_penalty=1.2,
|
|
return_full_text=True,
|
|
stop_sequences=["test"],
|
|
temperature=0.5,
|
|
top_p=0.9,
|
|
top_k=10,
|
|
truncate=5,
|
|
typical_p=0.9,
|
|
watermark=True,
|
|
decoder_input_details=True,
|
|
seed=0,
|
|
)
|
|
|
|
assert response.details.generated_tokens == 10
|
|
assert (
|
|
response.generated_text
|
|
== "What is gradient descent?\n\nHello! It seems you're addressing a"
|
|
)
|
|
assert response == response_snapshot
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_flash_phi35_moe_load(flash_phi35_moe, generate_load, response_snapshot):
|
|
responses = await generate_load(
|
|
flash_phi35_moe, "What is gradient descent?\n\n", max_new_tokens=10, n=4
|
|
)
|
|
|
|
assert len(responses) == 4
|
|
assert responses[0].details.generated_tokens == 10
|
|
assert (
|
|
responses[0].generated_text
|
|
== "Gradient descent is a first-order optimization algorithm"
|
|
)
|
|
assert all(
|
|
[r.generated_text == responses[0].generated_text for r in responses]
|
|
), f"{[r.generated_text for r in responses]}"
|
|
|
|
assert responses == response_snapshot
|