huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-09-11 20:34:54 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix: improve pali test and add snapshot

Browse Source
This commit is contained in:
drbh 2024-05-14 23:09:28 +00:00
parent d6e306c2b3
commit 70713fc292
2 changed files with 29 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

25
integration-tests/models/__snapshots__/test_flash_pali_gemma/test_flash_pali_gemma.json Normal file
View File

@ -0,0 +1,25 @@
{
"details": {
"best_of_sequences": null,
"finish_reason": "eos_token",
"generated_tokens": 2,
"prefill": [],
"seed": null,
"tokens": [
{
"id": 54901,
"logprob": -0.61621094,
"special": false,
"text": "beach"
},
{
"id": 1,
"logprob": -0.11273193,
"special": true,
"text": "<eos>"
}
],
"top_tokens": null
},
"generated_text": "beach"
}

7
integration-tests/models/test_flash_pali_gemma.py
View File

@ -7,7 +7,7 @@ import base64
@pytest.fixture(scope="module") @pytest.fixture(scope="module")
def flash_pali_gemma_handle(launcher): def flash_pali_gemma_handle(launcher):
with launcher( with launcher(
"Tinkering/test-bvhf", "gv-hf/paligemma-3b-mix-224",
num_shard=1, num_shard=1,
max_input_length=4000, max_input_length=4000,
max_total_tokens=4096, max_total_tokens=4096,
@ -31,7 +31,8 @@ def get_cow_beach():
@pytest.mark.private @pytest.mark.private
async def test_flash_pali_gemma(flash_pali_gemma, response_snapshot): async def test_flash_pali_gemma(flash_pali_gemma, response_snapshot):
cow = get_cow_beach() cow = get_cow_beach()
inputs = f"Where is the cow standing?\n![]({cow})" inputs = f"![]({cow})Where is the cow standing?\n"
response = await flash_pali_gemma.generate(inputs, max_new_tokens=20) response = await flash_pali_gemma.generate(inputs, max_new_tokens=20)
assert response.generated_text == "\nbeach" assert response.generated_text == "beach"
assert response == response_snapshot