liger-kernel-nightly 0.5.9.dev20250504194613__tar.gz → 0.5.9.dev20250505192105__tar.gz

{liger_kernel_nightly-0.5.9.dev20250504194613 → liger_kernel_nightly-0.5.9.dev20250505192105}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: liger_kernel_nightly
-Version: 0.5.9.dev20250504194613
+Version: 0.5.9.dev20250505192105
 Summary: Efficient Triton kernels for LLM Training
 License: BSD 2-CLAUSE LICENSE
         Copyright 2024 LinkedIn Corporation 

{liger_kernel_nightly-0.5.9.dev20250504194613 → liger_kernel_nightly-0.5.9.dev20250505192105}/benchmark/benchmarks_visualizer.py

@@ -123,7 +123,7 @@ def plot_data(df: pd.DataFrame, config: VisualizationsConfig):
     lines = ax.get_lines()
     colors = [line.get_color() for line in lines]
 
-    for (_, group_data), color in zip(df.groupby("kernel_provider"), colors, strict=False):
+    for (_, group_data), color in zip(df.groupby("kernel_provider"), colors):
         # for i, row in group_data.iterrows():
         y_error_lower = group_data["y_value_50"] - group_data["y_value_20"]
         y_error_upper = group_data["y_value_80"] - group_data["y_value_50"]

{liger_kernel_nightly-0.5.9.dev20250504194613 → liger_kernel_nightly-0.5.9.dev20250505192105}/examples/huggingface/training_multimodal.py

@@ -68,7 +68,7 @@ def construct_model_and_processor(model_name: str, use_liger: bool) -> torch.nn.
 def _validate_and_extract_the_cauldron(examples) -> dict[str, list]:
     batch_texts = []
     batch_images = []
-    for images, texts in zip(examples["images"], examples["texts"], strict=False):
+    for images, texts in zip(examples["images"], examples["texts"]):
         if not images:
             raise ValueError("No image found in example from the_cauldron dataset")
         if len(images) > 1:

{liger_kernel_nightly-0.5.9.dev20250504194613 → liger_kernel_nightly-0.5.9.dev20250505192105}/examples/medusa/train.py

@@ -175,7 +175,7 @@ def preprocess(
     input_ids = encoding.input_ids
 
     # Mask targets. Only compute loss on the assistant outputs.
-    for conv_index, (conversation, target, prompt) in enumerate(zip(conversations, targets, prompts, strict=False)):
+    for conv_index, (conversation, target, prompt) in enumerate(zip(conversations, targets, prompts)):
         # print(conv_index)
         for turn in conversation:
             if turn["role"] == "assistant":

{liger_kernel_nightly-0.5.9.dev20250504194613 → liger_kernel_nightly-0.5.9.dev20250505192105}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "liger_kernel_nightly"
-version = "0.5.9.dev20250504194613"
+version = "0.5.9.dev20250505192105"
 description = "Efficient Triton kernels for LLM Training"
 urls = { "Homepage" = "https://github.com/linkedin/Liger-Kernel" }
 readme = { file = "README.md", content-type = "text/markdown" }

{liger_kernel_nightly-0.5.9.dev20250504194613 → liger_kernel_nightly-0.5.9.dev20250505192105}/src/liger_kernel/chunked_loss/fused_linear_preference.py

@@ -222,7 +222,6 @@ class LigerFusedLinearPreferenceBase(torch.autograd.Function):
             (_ref_chosen_input_chunks if use_ref_model else [None] * len(_chosen_input_chunks)),
             (_ref_rejected_input_chunks if use_ref_model else [None] * len(_rejected_input_chunks)),
             (_chosen_nll_target_chunks if nll_target is not None else [None] * len(_chosen_input_chunks)),
-            strict=False,
         ):
             input_chunk = torch.cat([chosen_input_chunk, rejected_input_chunk], dim=0)
             ref_input_chunk = (

{liger_kernel_nightly-0.5.9.dev20250504194613 → liger_kernel_nightly-0.5.9.dev20250505192105}/src/liger_kernel_nightly.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: liger_kernel_nightly
-Version: 0.5.9.dev20250504194613
+Version: 0.5.9.dev20250505192105
 Summary: Efficient Triton kernels for LLM Training
 License: BSD 2-CLAUSE LICENSE
         Copyright 2024 LinkedIn Corporation 

{liger_kernel_nightly-0.5.9.dev20250504194613 → liger_kernel_nightly-0.5.9.dev20250505192105}/test/convergence/bf16/test_mini_models.py

@@ -1130,6 +1130,5 @@ def test_mini_model(
     for expected_param, actual_param in zip(
         expected_output["model"].named_parameters(),
         actual_output["model"].named_parameters(),
-        strict=False,
     ):
         assert_verbose_allclose(expected_param[1], actual_param[1], atol=param_atol, rtol=param_rtol)

{liger_kernel_nightly-0.5.9.dev20250504194613 → liger_kernel_nightly-0.5.9.dev20250505192105}/test/convergence/bf16/test_mini_models_multimodal.py

@@ -929,6 +929,5 @@ def test_mini_model_multimodal(
     for expected_param, actual_param in zip(
         expected_output["model"].named_parameters(),
         actual_output["model"].named_parameters(),
-        strict=False,
     ):
         assert_verbose_allclose(expected_param[1], actual_param[1], atol=param_atol, rtol=param_rtol)

{liger_kernel_nightly-0.5.9.dev20250504194613 → liger_kernel_nightly-0.5.9.dev20250505192105}/test/convergence/bf16/test_mini_models_with_logits.py

@@ -1129,6 +1129,5 @@ def test_mini_model(
     for expected_param, actual_param in zip(
         expected_output["model"].named_parameters(),
         actual_output["model"].named_parameters(),
-        strict=False,
     ):
         assert_verbose_allclose(expected_param[1], actual_param[1], atol=param_atol, rtol=param_rtol)

{liger_kernel_nightly-0.5.9.dev20250504194613 → liger_kernel_nightly-0.5.9.dev20250505192105}/test/convergence/fp32/test_mini_models.py

@@ -1000,6 +1000,5 @@ def test_mini_model(
     for expected_param, actual_param in zip(
         expected_output["model"].named_parameters(),
         actual_output["model"].named_parameters(),
-        strict=False,
     ):
         assert_verbose_allclose(expected_param[1], actual_param[1], atol=param_atol, rtol=param_rtol)

{liger_kernel_nightly-0.5.9.dev20250504194613 → liger_kernel_nightly-0.5.9.dev20250505192105}/test/convergence/fp32/test_mini_models_multimodal.py

@@ -910,6 +910,5 @@ def test_mini_model_multimodal(
     for expected_param, actual_param in zip(
         expected_output["model"].named_parameters(),
         actual_output["model"].named_parameters(),
-        strict=False,
     ):
         assert_verbose_allclose(expected_param[1], actual_param[1], atol=param_atol, rtol=param_rtol)

{liger_kernel_nightly-0.5.9.dev20250504194613 → liger_kernel_nightly-0.5.9.dev20250505192105}/test/convergence/fp32/test_mini_models_with_logits.py

@@ -1000,6 +1000,5 @@ def test_mini_model(
     for expected_param, actual_param in zip(
         expected_output["model"].named_parameters(),
         actual_output["model"].named_parameters(),
-        strict=False,
     ):
         assert_verbose_allclose(expected_param[1], actual_param[1], atol=param_atol, rtol=param_rtol)