fix incorrect loss value for vlms

Former-commit-id: 0aa29a71ce958343a2086090d647eb63b8f5f5be

requirements.txt

@@ -1,5 +1,5 @@
-transformers>=4.41.2,<=4.46.0
+transformers>=4.41.2,<=4.46.1
-datasets>=2.16.0,<=2.21.0
+datasets>=2.16.0,<=3.0.2
 accelerate>=0.34.0,<=1.0.1
 peft>=0.11.1,<=0.12.0
 trl>=0.8.6,<=0.9.6

src/llamafactory/__init__.py

@@ -20,17 +20,17 @@ Level:
 
 Dependency graph:
   main:
-    transformers>=4.41.2,<=4.46.0
+    transformers>=4.41.2,<=4.46.1
-    datasets>=2.16.0,<=2.21.0
+    datasets>=2.16.0,<=3.0.2
     accelerate>=0.34.0,<=1.0.1
     peft>=0.11.1,<=0.12.0
     trl>=0.8.6,<=0.9.6
   attention:
     transformers>=4.42.4 (gemma+fa2)
   longlora:
-    transformers>=4.41.2,<=4.46.0
+    transformers>=4.41.2,<=4.46.1
   packing:
-    transformers>=4.41.2,<=4.46.0
+    transformers>=4.41.2,<=4.46.1
 
 Disable version checking: DISABLE_VERSION_CHECK=1
 Enable VRAM recording: RECORD_VRAM=1

src/llamafactory/extras/misc.py

@@ -79,8 +79,8 @@ def check_dependencies() -> None:
     if os.environ.get("DISABLE_VERSION_CHECK", "0").lower() in ["true", "1"]:
         logger.warning("Version checking has been disabled, may lead to unexpected behaviors.")
     else:
-        require_version("transformers>=4.41.2,<=4.46.0", "To fix: pip install transformers>=4.41.2,<=4.46.0")
+        require_version("transformers>=4.41.2,<=4.46.1", "To fix: pip install transformers>=4.41.2,<=4.46.1")
-        require_version("datasets>=2.16.0,<=2.21.0", "To fix: pip install datasets>=2.16.0,<=2.21.0")
+        require_version("datasets>=2.16.0,<=3.0.2", "To fix: pip install datasets>=2.16.0,<=3.0.2")
         require_version("accelerate>=0.34.0,<=1.0.1", "To fix: pip install accelerate>=0.34.0,<=1.0.1")
         require_version("peft>=0.11.1,<=0.12.0", "To fix: pip install peft>=0.11.1,<=0.12.0")
         require_version("trl>=0.8.6,<=0.9.6", "To fix: pip install trl>=0.8.6,<=0.9.6")
@@ -237,7 +237,7 @@ def try_download_model_from_other_hub(model_args: "ModelArguments") -> str:
 
     if use_modelscope():
         require_version("modelscope>=1.11.0", "To fix: pip install modelscope>=1.11.0")
-        from modelscope import snapshot_download
+        from modelscope import snapshot_download  # type: ignore
 
         revision = "master" if model_args.model_revision == "main" else model_args.model_revision
         return snapshot_download(
@@ -248,7 +248,7 @@ def try_download_model_from_other_hub(model_args: "ModelArguments") -> str:
 
     if use_openmind():
         require_version("openmind>=0.8.0", "To fix: pip install openmind>=0.8.0")
-        from openmind.utils.hub import snapshot_download
+        from openmind.utils.hub import snapshot_download  # type: ignore
 
         return snapshot_download(
             model_args.model_name_or_path,

src/llamafactory/extras/packages.py

@@ -81,7 +81,7 @@ def is_transformers_version_greater_than_4_43():
 
 @lru_cache
 def is_transformers_version_equal_to_4_46():
-    return _get_package_version("transformers") == version.parse("4.46.0")
+    return version.parse("4.46.0") <= _get_package_version("transformers") <= version.parse("4.46.1")
 
 
 def is_uvicorn_available():

src/llamafactory/model/model_utils/longlora.py

@@ -353,7 +353,7 @@ def llama_sdpa_attention_forward(
 
 
 def _apply_llama_patch() -> None:
-    require_version("transformers>=4.41.2,<=4.46.0", "To fix: pip install transformers>=4.41.2,<=4.46.0")
+    require_version("transformers>=4.41.2,<=4.46.1", "To fix: pip install transformers>=4.41.2,<=4.46.1")
     LlamaAttention.forward = llama_attention_forward
     LlamaFlashAttention2.forward = llama_flash_attention_2_forward
     LlamaSdpaAttention.forward = llama_sdpa_attention_forward

src/llamafactory/model/model_utils/packing.py

@@ -114,7 +114,7 @@ def get_unpad_data(attention_mask: "torch.Tensor") -> Tuple["torch.Tensor", "tor
 
 
 def _patch_for_block_diag_attn(model_type: str) -> None:
-    require_version("transformers>=4.41.2,<=4.46.0", "To fix: pip install transformers>=4.41.2,<=4.46.0")
+    require_version("transformers>=4.41.2,<=4.46.1", "To fix: pip install transformers>=4.41.2,<=4.46.1")
     if is_transformers_version_greater_than_4_43():
         import transformers.modeling_flash_attention_utils
 

src/llamafactory/train/dpo/trainer.py

@@ -101,7 +101,7 @@ class CustomDPOTrainer(DPOTrainer):
             self.callback_handler.add_callback(PissaConvertCallback)
 
         if finetuning_args.use_badam:
-            from badam import BAdamCallback, clip_grad_norm_old_version
+            from badam import BAdamCallback, clip_grad_norm_old_version  # type: ignore
 
             self.accelerator.clip_grad_norm_ = MethodType(clip_grad_norm_old_version, self.accelerator)
             self.add_callback(BAdamCallback)
@@ -274,7 +274,7 @@ class CustomDPOTrainer(DPOTrainer):
         https://github.com/huggingface/transformers/blob/v4.46.0/src/transformers/trainer.py#L3605
         """
         loss = super().compute_loss(model, inputs, return_outputs)
-        if kwargs.pop("num_items_in_batch", False) and is_transformers_version_equal_to_4_46():
+        if is_transformers_version_equal_to_4_46() and kwargs.pop("num_items_in_batch", False):
             loss /= self.args.gradient_accumulation_steps
 
         return loss

src/llamafactory/train/kto/trainer.py

@@ -96,7 +96,7 @@ class CustomKTOTrainer(KTOTrainer):
             self.add_callback(SaveProcessorCallback(processor))
 
         if finetuning_args.use_badam:
-            from badam import BAdamCallback, clip_grad_norm_old_version
+            from badam import BAdamCallback, clip_grad_norm_old_version  # type: ignore
 
             self.accelerator.clip_grad_norm_ = MethodType(clip_grad_norm_old_version, self.accelerator)
             self.add_callback(BAdamCallback)
@@ -247,7 +247,7 @@ class CustomKTOTrainer(KTOTrainer):
         https://github.com/huggingface/transformers/blob/v4.46.0/src/transformers/trainer.py#L3605
         """
         loss = super().compute_loss(model, inputs, return_outputs)
-        if kwargs.pop("num_items_in_batch", False) and is_transformers_version_equal_to_4_46():
+        if is_transformers_version_equal_to_4_46() and kwargs.pop("num_items_in_batch", False):
             loss /= self.args.gradient_accumulation_steps
 
         return loss

src/llamafactory/train/ppo/trainer.py

@@ -181,7 +181,7 @@ class CustomPPOTrainer(PPOTrainer, Trainer):
             self.add_callback(SaveProcessorCallback(processor))
 
         if finetuning_args.use_badam:
-            from badam import BAdamCallback, clip_grad_norm_old_version
+            from badam import BAdamCallback, clip_grad_norm_old_version  # type: ignore
 
             self.accelerator.clip_grad_norm_ = MethodType(clip_grad_norm_old_version, self.accelerator)
             self.add_callback(BAdamCallback)

src/llamafactory/train/pt/trainer.py

@@ -19,6 +19,7 @@ from transformers import Trainer
 from typing_extensions import override
 
 from ...extras.logging import get_logger
+from ...extras.packages import is_transformers_version_equal_to_4_46
 from ..callbacks import PissaConvertCallback, SaveProcessorCallback
 from ..trainer_utils import create_custom_optimizer, create_custom_scheduler
 
@@ -51,7 +52,7 @@ class CustomTrainer(Trainer):
             self.add_callback(PissaConvertCallback)
 
         if finetuning_args.use_badam:
-            from badam import BAdamCallback, clip_grad_norm_old_version
+            from badam import BAdamCallback, clip_grad_norm_old_version  # type: ignore
 
             self.accelerator.clip_grad_norm_ = MethodType(clip_grad_norm_old_version, self.accelerator)
             self.add_callback(BAdamCallback)
@@ -68,3 +69,15 @@ class CustomTrainer(Trainer):
     ) -> "torch.optim.lr_scheduler.LRScheduler":
         create_custom_scheduler(self.args, num_training_steps, optimizer)
         return super().create_scheduler(num_training_steps, optimizer)
+
+    @override
+    def compute_loss(self, model, inputs, return_outputs=False, **kwargs):
+        r"""
+        Fixes the loss value for transformers 4.46.0.
+        https://github.com/huggingface/transformers/blob/v4.46.0/src/transformers/trainer.py#L3605
+        """
+        loss = super().compute_loss(model, inputs, return_outputs, **kwargs)
+        if is_transformers_version_equal_to_4_46() and not getattr(self, "model_accepts_loss_kwargs", False):
+            loss /= self.args.gradient_accumulation_steps  # other model should not scale the loss
+
+        return loss

src/llamafactory/train/rm/trainer.py

@@ -60,7 +60,7 @@ class PairwiseTrainer(Trainer):
             self.add_callback(PissaConvertCallback)
 
         if finetuning_args.use_badam:
-            from badam import BAdamCallback, clip_grad_norm_old_version
+            from badam import BAdamCallback, clip_grad_norm_old_version  # type: ignore
 
             self.accelerator.clip_grad_norm_ = MethodType(clip_grad_norm_old_version, self.accelerator)
             self.add_callback(BAdamCallback)
@@ -100,7 +100,7 @@ class PairwiseTrainer(Trainer):
 
         loss = -torch.nn.functional.logsigmoid(chosen_scores.float() - rejected_scores.float()).mean()
 
-        if kwargs.pop("num_items_in_batch", False) and is_transformers_version_equal_to_4_46():
+        if is_transformers_version_equal_to_4_46() and kwargs.pop("num_items_in_batch", False):
             loss /= self.args.gradient_accumulation_steps  # fixes the loss value for transformers 4.46.0
 
         if return_outputs:

src/llamafactory/train/sft/trainer.py

@@ -27,6 +27,7 @@ from typing_extensions import override
 
 from ...extras.constants import IGNORE_INDEX
 from ...extras.logging import get_logger
+from ...extras.packages import is_transformers_version_equal_to_4_46
 from ..callbacks import PissaConvertCallback, SaveProcessorCallback
 from ..trainer_utils import create_custom_optimizer, create_custom_scheduler
 
@@ -60,7 +61,7 @@ class CustomSeq2SeqTrainer(Seq2SeqTrainer):
             self.add_callback(PissaConvertCallback)
 
         if finetuning_args.use_badam:
-            from badam import BAdamCallback, clip_grad_norm_old_version
+            from badam import BAdamCallback, clip_grad_norm_old_version  # type: ignore
 
             self.accelerator.clip_grad_norm_ = MethodType(clip_grad_norm_old_version, self.accelerator)
             self.add_callback(BAdamCallback)
@@ -78,6 +79,18 @@ class CustomSeq2SeqTrainer(Seq2SeqTrainer):
         create_custom_scheduler(self.args, num_training_steps, optimizer)
         return super().create_scheduler(num_training_steps, optimizer)
 
+    @override
+    def compute_loss(self, model, inputs, return_outputs=False, **kwargs):
+        r"""
+        Fixes the loss value for transformers 4.46.0.
+        https://github.com/huggingface/transformers/blob/v4.46.0/src/transformers/trainer.py#L3605
+        """
+        loss = super().compute_loss(model, inputs, return_outputs, **kwargs)
+        if is_transformers_version_equal_to_4_46() and not getattr(self, "model_accepts_loss_kwargs", False):
+            loss /= self.args.gradient_accumulation_steps  # other model should not scale the loss
+
+        return loss
+
     @override
     def prediction_step(
         self,