tocmo0nlord/axolotl
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

make sure the all of the model is on the same device, so this test will pass on multigpu (#2524) [skip ci]

Browse Source
This commit is contained in:
Wing Lian
2025-04-15 22:15:42 -07:00
committed by GitHub
parent e4307fb7d7
commit 198d775d6d
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
tests/e2e/patched/lora_kernels/test_lora_kernel_patching.py
View File

@@ -144,7 +144,7 @@ def test_swiglu_mlp_integration(small_llama_model):
def test_geglu_model_integration():
"""Test GeGLU activation with Gemma model."""
model = AutoModelForCausalLM.from_pretrained(
"mhenrichsen/gemma-2b", torch_dtype=torch.float16, device_map="auto"
"mhenrichsen/gemma-2b", torch_dtype=torch.float16, device_map="cuda:0"
)
peft_config = get_peft_config(
{
@@ -347,7 +347,7 @@ def test_model_architecture(model_config):
"""Test LoRA kernel patches across different model architectures."""
# Load model with appropriate dtype
model = AutoModelForCausalLM.from_pretrained(
model_config["name"], torch_dtype=model_config["dtype"], device_map="auto"
model_config["name"], torch_dtype=model_config["dtype"], device_map="cuda:0"
)
# Apply LoRA configuration