huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-06-14 13:22:07 +00:00
Code Issues Packages Projects Releases Wiki Activity

tests(neuron): remove obsolete models

Browse Source
This commit is contained in:
David Corvoysier 2025-05-26 13:54:41 +00:00
parent 03b9004fde
commit 82f37d3f60
2 changed files with 0 additions and 22 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
integration-tests/fixtures/neuron/export_models.py
View File

@ -28,15 +28,6 @@ logger = logging.getLogger(__file__)
# All model configurations below will be added to the neuron_model_config fixture
MODEL_CONFIGURATIONS = {
"gpt2": {
"model_id": "gpt2",
"export_kwargs": {
"batch_size": 4,
"sequence_length": 1024,
"num_cores": 2,
"auto_cast_type": "fp16",
},
},
"llama": {
"model_id": "unsloth/Llama-3.2-1B-Instruct",
"export_kwargs": {
@ -46,15 +37,6 @@ MODEL_CONFIGURATIONS = {
"auto_cast_type": "fp16",
},
},
"mistral": {
"model_id": "optimum/mistral-1.1b-testing",
"export_kwargs": {
"batch_size": 4,
"sequence_length": 4096,
"num_cores": 2,
"auto_cast_type": "bf16",
},
},
"qwen2": {
"model_id": "Qwen/Qwen2.5-0.5B",
"export_kwargs": {

4
integration-tests/neuron/test_generate.py
View File

@ -20,9 +20,7 @@ async def test_model_single_request(tgi_service):
)
assert response.details.generated_tokens == 17
greedy_expectations = {
"gpt2": "\n\nDeep learning is a new field of research that has been around for a while",
"llama": " and How Does it Work?\nDeep learning is a subset of machine learning that uses artificial",
"mistral": "\nWhat is Deep Learning?\nDeep Learning is a type of machine learning that",
"qwen2": " - Part 1\n\nDeep Learning is a subset of Machine Learning that is based on",
"granite": "\n\nDeep Learning is a subset of Machine Learning, which is a branch of Art",
}
@ -79,9 +77,7 @@ async def test_model_multiple_requests(tgi_service, neuron_generate_load):
assert len(responses) == 4
expectations = {
"gpt2": "Deep learning is a new field of research that has been around for a while",
"llama": "Deep learning is a subset of machine learning that uses artificial",
"mistral": "Deep Learning is a type of machine learning that",
"qwen2": "Deep Learning is a subset of Machine Learning that is based on",
"granite": "Deep Learning is a subset of Machine Learning, which is a branch of Art",
}