Merge pull request #1343 from Hypothesis-Z/fix-transformers-4.48.0

fix transformers 4.48.0
FlagOpen · Feb 7, 2025 · 42a25b1 · 42a25b1
2 parents 0082bcf + 99dfb3d
commit 42a25b1
Show file tree

Hide file tree

Showing 3 changed files with 6 additions and 9 deletions.
diff --git a/FlagEmbedding/inference/reranker/decoder_only/models/gemma_model.py b/FlagEmbedding/inference/reranker/decoder_only/models/gemma_model.py
@@ -53,7 +53,7 @@
 )
 from .gemma_config import CostWiseGemmaConfig
 from transformers.models.gemma2.modeling_gemma2 import Gemma2RMSNorm, Gemma2RotaryEmbedding, rotate_half, apply_rotary_pos_emb
-from transformers.models.gemma2.modeling_gemma2 import Gemma2MLP, repeat_kv, Gemma2Attention, Gemma2FlashAttention2, Gemma2SdpaAttention, GEMMA2_ATTENTION_CLASSES, Gemma2DecoderLayer, GEMMA2_START_DOCSTRING
+from transformers.models.gemma2.modeling_gemma2 import Gemma2MLP, repeat_kv, Gemma2Attention, Gemma2DecoderLayer, GEMMA2_START_DOCSTRING
 from transformers.models.gemma2.modeling_gemma2 import GEMMA2_INPUTS_DOCSTRING
 
 if is_flash_attn_2_available():
@@ -105,12 +105,6 @@ def _init_weights(self, module):
             if module.padding_idx is not None:
                 module.weight.data[module.padding_idx].zero_()
 
-GEMMA2_ATTENTION_CLASSES = {
-    "eager": Gemma2Attention,
-    "flash_attention_2": Gemma2FlashAttention2,
-    "sdpa": Gemma2SdpaAttention,
-}
-
 
 _CONFIG_FOR_DOC = "CostWiseGemmaConfig"
 

diff --git a/FlagEmbedding/inference/reranker/decoder_only/models/modeling_minicpm_reranker.py b/FlagEmbedding/inference/reranker/decoder_only/models/modeling_minicpm_reranker.py
@@ -41,7 +41,7 @@
 from transformers.modeling_outputs import BaseModelOutputWithPast, CausalLMOutputWithPast, \
     SequenceClassifierOutputWithPast
 from transformers.modeling_utils import PreTrainedModel
-from transformers.pytorch_utils import ALL_LAYERNORM_LAYERS, is_torch_greater_or_equal_than_1_13
+from transformers.pytorch_utils import ALL_LAYERNORM_LAYERS
 from transformers.utils import (
     add_start_docstrings,
     add_start_docstrings_to_model_forward,
@@ -63,6 +63,9 @@
 
 # This makes `_prepare_4d_causal_attention_mask` a leaf function in the FX graph.
 # It means that the function will not be traced through and simply appear as a node in the graph.
+from packaging import version
+parsed_torch_version_base = version.parse(version.parse(torch.__version__).base_version)
+is_torch_greater_or_equal_than_1_13 = parsed_torch_version_base >= version.parse("1.13")
 if is_torch_fx_available():
     if not is_torch_greater_or_equal_than_1_13:
         import torch.fx

diff --git a/setup.py b/setup.py
@@ -15,7 +15,7 @@
     include_package_data=True,
     install_requires=[
         'torch>=1.6.0',
-        'transformers==4.44.2',
+        'transformers>=4.44.2',
         'datasets==2.19.0',
         'accelerate>=0.20.1',
         'sentence_transformers',