diff --git a/tests/test_trainers_args.py b/tests/test_trainers_args.py
index 3aff81fd3f..25ed71ff0b 100644
--- a/tests/test_trainers_args.py
+++ b/tests/test_trainers_args.py
@@ -153,7 +153,6 @@ def test_dpo(self):
                 max_length=256,
                 max_prompt_length=64,
                 max_completion_length=64,
-                is_encoder_decoder=True,
                 disable_dropout=False,
                 # generate_during_eval=True, # ignore this one, it requires wandb
                 precompute_ref_log_probs=True,
@@ -188,7 +187,6 @@ def test_dpo(self):
             self.assertEqual(trainer.args.max_length, 256)
             self.assertEqual(trainer.args.max_prompt_length, 64)
             self.assertEqual(trainer.args.max_completion_length, 64)
-            self.assertEqual(trainer.args.is_encoder_decoder, True)
             self.assertEqual(trainer.args.disable_dropout, False)
             # self.assertEqual(trainer.args.generate_during_eval, True)
             self.assertEqual(trainer.args.precompute_ref_log_probs, True)
diff --git a/trl/trainer/dpo_config.py b/trl/trainer/dpo_config.py
index a3cdc28d28..55c6ecc7c8 100644
--- a/trl/trainer/dpo_config.py
+++ b/trl/trainer/dpo_config.py
@@ -12,7 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import warnings
 from dataclasses import dataclass, field
 from enum import Enum
 from typing import Any, Callable, Optional, Union
@@ -385,18 +384,3 @@ class DPOConfig(TrainingArguments):
             "Comet during evaluation."
         },
     )
-
-    # Deprecated parameters
-    is_encoder_decoder: Optional[bool] = field(
-        default=None,
-        metadata={"help": "Deprecated. This argument is not used anymore."},
-    )
-
-    def __post_init__(self):
-        if self.is_encoder_decoder is not None:
-            warnings.warn(
-                "The `is_encoder_decoder` parameter is deprecated will be removed in version 0.15. The trainer now "
-                "automatically determines if the model is an encoder-decoder, so you can safely remove it."
-            )
-
-        return super().__post_init__()
diff --git a/trl/trainer/orpo_trainer.py b/trl/trainer/orpo_trainer.py
index 06457683e8..832927d05c 100644
--- a/trl/trainer/orpo_trainer.py
+++ b/trl/trainer/orpo_trainer.py
@@ -50,7 +50,6 @@
 from transformers.trainer_callback import TrainerCallback
 from transformers.trainer_utils import EvalLoopOutput
 from transformers.utils import is_peft_available, is_torch_fx_proxy
-from transformers.utils.deprecation import deprecate_kwarg
 
 from ..data_utils import maybe_apply_chat_template, maybe_extract_prompt
 from ..models import PreTrainedModelWrapper
@@ -119,9 +118,6 @@ class ORPOTrainer(Trainer):
 
     _tag_names = ["trl", "orpo"]
 
-    @deprecate_kwarg(
-        "tokenizer", "0.15.0", "processing_class", warn_if_greater_or_equal_version=True, raise_if_both_names=True
-    )
     def __init__(
         self,
         model: Optional[Union[PreTrainedModel, nn.Module, str]] = None,
diff --git a/trl/trainer/ppo_trainer.py b/trl/trainer/ppo_trainer.py
index 83926cfd6a..27cbdd016c 100644
--- a/trl/trainer/ppo_trainer.py
+++ b/trl/trainer/ppo_trainer.py
@@ -46,7 +46,6 @@
 from transformers.trainer import DEFAULT_CALLBACKS, DEFAULT_PROGRESS_CALLBACK
 from transformers.trainer_callback import CallbackHandler, ExportableState, PrinterCallback
 from transformers.utils import is_peft_available
-from transformers.utils.deprecation import deprecate_kwarg
 
 from ..core import masked_mean, masked_whiten
 from ..models import create_reference_model
@@ -98,14 +97,6 @@ def forward(self, **kwargs):
 class PPOTrainer(Trainer):
     _tag_names = ["trl", "ppo"]
 
-    @deprecate_kwarg("config", "0.15.0", "args", warn_if_greater_or_equal_version=True, raise_if_both_names=True)
-    @deprecate_kwarg(
-        "tokenizer", "0.15.0", "processing_class", warn_if_greater_or_equal_version=True, raise_if_both_names=True
-    )
-    @deprecate_kwarg("policy", "0.15.0", "model", warn_if_greater_or_equal_version=True, raise_if_both_names=True)
-    @deprecate_kwarg(
-        "ref_policy", "0.15.0", "ref_model", warn_if_greater_or_equal_version=True, raise_if_both_names=True
-    )
     def __init__(
         self,
         args: PPOConfig,
diff --git a/trl/trainer/reward_trainer.py b/trl/trainer/reward_trainer.py
index ea25b425ab..063fe5e8e8 100644
--- a/trl/trainer/reward_trainer.py
+++ b/trl/trainer/reward_trainer.py
@@ -39,7 +39,6 @@
 from transformers.trainer_pt_utils import nested_detach
 from transformers.trainer_utils import EvalPrediction
 from transformers.utils import is_peft_available
-from transformers.utils.deprecation import deprecate_kwarg
 
 from ..data_utils import maybe_apply_chat_template
 from .reward_config import RewardConfig
@@ -84,9 +83,6 @@ def _tokenize(batch: dict[str, list[Any]], tokenizer: "PreTrainedTokenizerBase")
 class RewardTrainer(Trainer):
     _tag_names = ["trl", "reward-trainer"]
 
-    @deprecate_kwarg(
-        "tokenizer", "0.15.0", "processing_class", warn_if_greater_or_equal_version=True, raise_if_both_names=True
-    )
     def __init__(
         self,
         model: Optional[Union[PreTrainedModel, nn.Module]] = None,