mirror of
https://github.com/huggingface/text-generation-inference.git
synced 2025-04-20 22:32:07 +00:00
a6a0c97ed9
* wip * rollback * refactor to use prefix/postfix namming + fix all_input_ids_tensor * maybe patching vlms? * fix filter and concat * wip, no filter, no concat * current * add prepare_for_prefill * working * load tested * re-create slots * re-create slots * fix slot_filtering_indices * feedback loop * remove log * fix benchmarker * fix vlm and seq2seq * rename to cache and input lengths * fix prefill logprobs * fix launcher * fix logprobs? * idk at this point * max input length * omfg * remove debugging lines * fix tests * fix mllama * fix cargo tests * remove support chunking for paged * Fixing non blocked attentions * Fixing dtype + AMD, Ipex targets. * lint fix. * rename * Fix prefix_caching variable, remove defaults in server (confusing a lot of the times). * Add simple resolution when user specifies ATTENTION=paged. * Put back non default simple tests. * Fix env name --------- Co-authored-by: Nicolas Patry <patry.nicolas@protonmail.com>
66 lines
1.9 KiB
Python
66 lines
1.9 KiB
Python
import pytest
|
|
|
|
|
|
@pytest.fixture(scope="module")
|
|
def idefics_handle(launcher):
|
|
with launcher(
|
|
"HuggingFaceM4/idefics-9b-instruct", num_shard=2, dtype="float16"
|
|
) as handle:
|
|
yield handle
|
|
|
|
|
|
@pytest.fixture(scope="module")
|
|
async def idefics(idefics_handle):
|
|
await idefics_handle.health(300)
|
|
return idefics_handle.client
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_idefics(idefics, response_snapshot, chicken):
|
|
response = await idefics.generate(
|
|
f"User:Can you tell me a very short story based on the image?",
|
|
max_new_tokens=10,
|
|
decoder_input_details=True,
|
|
)
|
|
|
|
assert response.details.generated_tokens == 10
|
|
assert (
|
|
response.generated_text == " \nAssistant: A rooster stands"
|
|
), f"{repr(response.generated_text)}"
|
|
assert response == response_snapshot
|
|
|
|
|
|
@pytest.mark.release
|
|
@pytest.mark.asyncio
|
|
@pytest.mark.private
|
|
async def test_idefics_two_images(idefics, response_snapshot, chicken, cow_beach):
|
|
response = await idefics.generate(
|
|
f"User:Where are the cow and chicken?<end_of_utterance> \nAssistant:",
|
|
max_new_tokens=20,
|
|
)
|
|
assert (
|
|
response.generated_text == " The cow and chicken are on a beach."
|
|
), f"{repr(response.generated_text)}"
|
|
assert response == response_snapshot
|
|
|
|
|
|
@pytest.mark.release
|
|
@pytest.mark.asyncio
|
|
async def test_idefics_load(idefics, generate_load, response_snapshot, chicken):
|
|
responses = await generate_load(
|
|
idefics,
|
|
f"User:Can you tell me a very short story based on the image?",
|
|
max_new_tokens=10,
|
|
n=4,
|
|
)
|
|
|
|
generated_texts = [r.generated_text for r in responses]
|
|
|
|
assert generated_texts[0] == " \nAssistant: A rooster stands"
|
|
assert len(generated_texts) == 4
|
|
assert generated_texts, all(
|
|
[text == generated_texts[0] for text in generated_texts]
|
|
)
|
|
|
|
assert responses == response_snapshot
|