SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Remove all traces of low_cpu_mem_usage (#38792)

Browse Source 
* remove it from all py files

* remove it from the doc

* remove it from examples

* style

* remove traces of _fast_init

* Update test_peft_integration.py

* CIs
This commit is contained in:
Cyril Vallez
2025-06-12 16:39:33 +02:00
committed by GitHub
parent 3542e0b844
commit 4b8ec667e9
76 changed files with 100 additions and 598 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
tests/models/glm4/test_modeling_glm4.py
View File

@@ -104,9 +104,7 @@ class Glm4IntegrationTest(unittest.TestCase):
)
EXPECTED_TEXT = EXPECTED_TEXTS.get_expectation()
model = AutoModelForCausalLM.from_pretrained(
self.model_id, low_cpu_mem_usage=True, torch_dtype=torch.float16
).to(torch_device)
model = AutoModelForCausalLM.from_pretrained(self.model_id, torch_dtype=torch.float16).to(torch_device)
tokenizer = AutoTokenizer.from_pretrained(self.model_id)
inputs = tokenizer(self.input_text, return_tensors="pt", padding=True).to(torch_device)
@@ -132,9 +130,7 @@ class Glm4IntegrationTest(unittest.TestCase):
)
EXPECTED_TEXT = EXPECTED_TEXTS.get_expectation()
model = AutoModelForCausalLM.from_pretrained(
self.model_id, low_cpu_mem_usage=True, torch_dtype=torch.bfloat16
).to(torch_device)
model = AutoModelForCausalLM.from_pretrained(self.model_id, torch_dtype=torch.bfloat16).to(torch_device)
tokenizer = AutoTokenizer.from_pretrained(self.model_id)
inputs = tokenizer(self.input_text, return_tensors="pt", padding=True).to(torch_device)
@@ -162,7 +158,6 @@ class Glm4IntegrationTest(unittest.TestCase):
model = AutoModelForCausalLM.from_pretrained(
self.model_id,
low_cpu_mem_usage=True,
torch_dtype=torch.bfloat16,
attn_implementation="eager",
)
@@ -195,7 +190,6 @@ class Glm4IntegrationTest(unittest.TestCase):
model = AutoModelForCausalLM.from_pretrained(
self.model_id,
low_cpu_mem_usage=True,
torch_dtype=torch.bfloat16,
attn_implementation="sdpa",
)
@@ -226,7 +220,6 @@ class Glm4IntegrationTest(unittest.TestCase):
model = AutoModelForCausalLM.from_pretrained(
self.model_id,
low_cpu_mem_usage=True,
torch_dtype=torch.bfloat16,
attn_implementation="flash_attention_2",
)