fix(deps): breaking change from transformers 5.4.0 (#3231)

* fix(deps): breaking change from transformers 5.4.0

* Update src/lerobot/policies/xvla/modeling_florence2.py

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Signed-off-by: Maxime Ellerbach <maxime@ellerbach.net>

* Update src/lerobot/policies/wall_x/qwen_model/qwen2_5_vl_moe.py

Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Signed-off-by: Maxime Ellerbach <maxime@ellerbach.net>

* removing dataclass

* bumping transformers 5.4.0

---------

Signed-off-by: Maxime Ellerbach <maxime@ellerbach.net>
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>

src/lerobot/policies/groot/action_head/flow_matching_action_head.py

@@ -13,7 +13,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from dataclasses import dataclass, field
+from dataclasses import field
 from typing import TYPE_CHECKING
 
 import torch
@@ -110,7 +110,6 @@ class MultiEmbodimentActionEncoder(nn.Module):
         return x
 
 
-@dataclass
 class FlowmatchingActionHeadConfig(PretrainedConfig):
     """NOTE: N1.5 uses XEmbFlowmatchingPolicyHeadConfig as action head"""
 

src/lerobot/policies/groot/groot_n1.py

@@ -13,7 +13,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from dataclasses import dataclass, field
+from dataclasses import field
 from pathlib import Path
 from typing import TYPE_CHECKING
 
@@ -173,7 +173,6 @@ N_COLOR_CHANNELS = 3
 
 
 # config
-@dataclass
 class GR00TN15Config(PretrainedConfig):
     model_type = "gr00t_n1_5"
     backbone_cfg: dict = field(init=False, metadata={"help": "Backbone configuration."})

src/lerobot/policies/wall_x/qwen_model/qwen2_5_vl_moe.py

@@ -22,7 +22,7 @@ from transformers.utils import (
     add_start_docstrings,
     add_start_docstrings_to_model_forward,
     is_flash_attn_2_available,
-    is_flash_attn_greater_or_equal_2_10,
+    is_flash_attn_greater_or_equal,
     is_torchdynamo_compiling,
     logging,
     replace_return_docstrings,
@@ -890,7 +890,7 @@ class Qwen2_5_VLFlashAttention2(Qwen2_5_VLAttention):
         # TODO: Should be removed once Flash Attention for RoCm is bumped to 2.1.
         # flash_attn<2.1 generates top-left aligned causal mask, while what is needed here is bottom-right alignment, that was made default for flash_attn>=2.1. This attribute is used to handle this difference. Reference: https://github.com/Dao-AILab/flash-attention/releases/tag/v2.1.0.
         # Beware that with flash_attn<2.1, using q_seqlen != k_seqlen (except for the case q_seqlen == 1) produces a wrong mask (top-left).
-        self._flash_attn_uses_top_left_mask = not is_flash_attn_greater_or_equal_2_10()
+        self._flash_attn_uses_top_left_mask = not is_flash_attn_greater_or_equal("2.1.0")
 
     def forward(
         self,

src/lerobot/policies/xvla/modeling_florence2.py

@@ -45,7 +45,7 @@ from transformers.utils import (
     add_start_docstrings,
     add_start_docstrings_to_model_forward,
     is_flash_attn_2_available,
-    is_flash_attn_greater_or_equal_2_10,
+    is_flash_attn_greater_or_equal,
     logging,
     replace_return_docstrings,
 )
@@ -909,7 +909,7 @@ class Florence2FlashAttention2(Florence2Attention):
         # TODO: Should be removed once Flash Attention for RoCm is bumped to 2.1.
         # flash_attn<2.1 generates top-left aligned causal mask, while what is needed here is bottom-right alignment, that was made default for flash_attn>=2.1. This attribute is used to handle this difference. Reference: https://github.com/Dao-AILab/flash-attention/releases/tag/v2.1.0.
         # Beware that with flash_attn<2.1, using q_seqlen != k_seqlen (except for the case q_seqlen == 1) produces a wrong mask (top-left).
-        self._flash_attn_uses_top_left_mask = not is_flash_attn_greater_or_equal_2_10()
+        self._flash_attn_uses_top_left_mask = not is_flash_attn_greater_or_equal("2.1.0")
 
     def _reshape(self, tensor: torch.Tensor, seq_len: int, bsz: int):
         return tensor.view(bsz, seq_len, self.num_heads, self.head_dim)