Bump transformers 4.53.0 (#3618)

AlexanderDokuchaev · web-flow · commit a8df402f1640 · 2025-10-29T14:49:41.000+02:00
### Changes Bump packages ``` transformers==4.53.0 optimum-intel==1.26.0 optimum==2.0.0 ``` ### Tests WC: https://github.com/openvinotoolkit/nncf/actions/runs/18844019917 Examples; https://github.com/openvinotoolkit/nncf/actions/runs/18905470590 PTQ: manual/job/post_training_quantization/741/ Nightly Torch: nightly/job/torch_nightly/693/
diff --git a/examples/llm_compression/onnx/tiny_llama/main.py b/examples/llm_compression/onnx/tiny_llama/main.py
@@ -27,6 +27,11 @@
 MODEL_ID = "TinyLlama/TinyLlama-1.1B-Chat-v1.0"
 OUTPUT_DIR = ROOT / "tinyllama_compressed"
 
+# TODO(AlexanderDokuchaev): WA for https://github.com/huggingface/optimum-intel/issues/1498
+from optimum.exporters.tasks import TasksManager  # noqa: E402
+
+TasksManager._TRANSFORMERS_TASKS_TO_MODEL_LOADERS["image-text-to-text"] = "AutoModelForImageTextToText"
+
 
 def main():
     # Export the pretrained model in ONNX format. The OUTPUT_DIR directory
diff --git a/examples/llm_compression/onnx/tiny_llama/requirements.txt b/examples/llm_compression/onnx/tiny_llama/requirements.txt
@@ -1,6 +1,8 @@
-transformers==4.52.1
+transformers==4.53.0
 openvino==2025.3.0
-optimum-intel[openvino]
-git+https://github.com/onnx/onnx.git@c25eebcf51b781dbfcc75a9c8bdf5dd1781367fe # onnx-1.19.0.dev
+optimum-intel[openvino]==1.26.0
+optimum-onnx==0.0.3
+optimum==2.0.0
+onnx==1.19.1
 onnxruntime==1.21.1
-torch==2.8.0
+torch==2.9.0
diff --git a/examples/llm_compression/onnx/tiny_llama_scale_estimation/main.py b/examples/llm_compression/onnx/tiny_llama_scale_estimation/main.py
@@ -34,6 +34,11 @@
 warnings.filterwarnings("ignore", category=torch.jit.TracerWarning)
 warnings.filterwarnings("ignore", category=OnnxExporterWarning)
 
+# TODO(AlexanderDokuchaev): WA for https://github.com/huggingface/optimum-intel/issues/1498
+from optimum.exporters.tasks import TasksManager  # noqa: E402
+
+TasksManager._TRANSFORMERS_TASKS_TO_MODEL_LOADERS["image-text-to-text"] = "AutoModelForImageTextToText"
+
 
 def tiny_llama_transform_func(
     item: dict[str, str], tokenizer: LlamaTokenizerFast, onnx_model: onnx.ModelProto
diff --git a/examples/llm_compression/onnx/tiny_llama_scale_estimation/requirements.txt b/examples/llm_compression/onnx/tiny_llama_scale_estimation/requirements.txt
@@ -1,7 +1,9 @@
-torch==2.8.0
-transformers==4.52.1
+torch==2.9.0
+transformers==4.53.0
 openvino==2025.3.0
-optimum-intel[openvino]
+optimum-intel[openvino]==1.26.0
+optimum-onnx==0.0.3
+optimum==2.0.0
 onnx==1.17.0
 onnxruntime==1.21.1
 datasets==2.14.7
diff --git a/examples/llm_compression/openvino/smollm2_360m_codebook/requirements.txt b/examples/llm_compression/openvino/smollm2_360m_codebook/requirements.txt
@@ -1,6 +1,8 @@
 openvino==2025.3.0
-optimum-intel[openvino]>=1.22.0
-transformers==4.52.1
+optimum-intel[openvino]==1.26.0
+optimum-onnx==0.0.3
+optimum==2.0.0
+transformers==4.53.0
 onnx==1.17.0
-torch==2.8.0
-torchvision==0.23.0
+torch==2.9.0
+torchvision==0.24.0
diff --git a/examples/llm_compression/openvino/smollm2_360m_fp8/requirements.txt b/examples/llm_compression/openvino/smollm2_360m_fp8/requirements.txt
@@ -1,6 +1,8 @@
-datasets
+datasets==4.3.0
 openvino==2025.3.0
-optimum-intel[openvino]>=1.22.0
-transformers==4.52.1
+optimum-intel[openvino]==1.26.0
+optimum-onnx==0.0.3
+optimum==2.0.0
+transformers==4.53.0
 onnx==1.17.0
-torch==2.8.0
+torch==2.9.0
diff --git a/examples/llm_compression/openvino/tiny_llama/requirements.txt b/examples/llm_compression/openvino/tiny_llama/requirements.txt
@@ -1,6 +1,8 @@
-transformers==4.52.1
 datasets==2.14.7
-openvino==2025.3.0
-optimum-intel[openvino]>=1.22.0
 onnx==1.17.0
-torch==2.8.0
+openvino==2025.3.0
+optimum-intel[openvino]==1.26.0
+optimum-onnx==0.0.3
+optimum==2.0.0
+torch==2.9.0
+transformers==4.53.0
diff --git a/examples/llm_compression/openvino/tiny_llama_find_hyperparams/requirements.txt b/examples/llm_compression/openvino/tiny_llama_find_hyperparams/requirements.txt
@@ -1,7 +1,9 @@
 whowhatbench @ git+https://github.com/openvinotoolkit/openvino.genai@2025.3.0.0#subdirectory=tools/who_what_benchmark
 numpy==1.26.4
 openvino==2025.3.0
-optimum-intel==1.24.0
-transformers==4.52.1
+optimum-intel==1.26.0
+optimum-onnx==0.0.3
+optimum==2.0.0
+transformers==4.53.0
 onnx==1.17.0
 torch==2.9.0
diff --git a/examples/llm_compression/openvino/tiny_llama_synthetic_data/requirements.txt b/examples/llm_compression/openvino/tiny_llama_synthetic_data/requirements.txt
@@ -1,7 +1,9 @@
-torch==2.8.0
+torch==2.9.0
 datasets==3.0.1
 numpy>=1.23.5,<2
 openvino==2025.3.0
-optimum-intel>=1.22.0
-transformers==4.52.1
+optimum-intel==1.26.0
+optimum-onnx==0.0.3
+optimum==2.0.0
+transformers==4.53.0
 onnx==1.17.0
diff --git a/examples/llm_compression/torch/distillation_qat_with_lora/requirements.txt b/examples/llm_compression/torch/distillation_qat_with_lora/requirements.txt
@@ -1,7 +1,9 @@
 tensorboard==2.13.0
-torch==2.8.0
+torch==2.9.0
 numpy>=1.23.5,<2
 openvino==2025.3.0
-optimum-intel>=1.22.0
-transformers==4.52.1
+optimum-intel==1.26.0
+optimum-onnx==0.0.3
+optimum==2.0.0
+transformers==4.53.0
 lm_eval==0.4.8
diff --git a/examples/llm_compression/torch/downstream_qat_with_nls/requirements.txt b/examples/llm_compression/torch/downstream_qat_with_nls/requirements.txt
@@ -1,7 +1,9 @@
 tensorboard==2.13.0
-torch==2.8.0
+torch==2.9.0
 numpy>=1.23.5,<2
 openvino==2025.3.0
-optimum-intel>=1.22.0
-transformers==4.52.1
+optimum-intel==1.26.0
+optimum-onnx==0.0.3
+optimum==2.0.0
+transformers==4.53.0
 lm_eval==0.4.8
diff --git a/examples/llm_compression/torch_fx/tiny_llama/requirements.txt b/examples/llm_compression/torch_fx/tiny_llama/requirements.txt
@@ -1,6 +1,6 @@
-transformers==4.52.1
+transformers==4.53.0
 datasets==2.14.7
 openvino==2025.3.0
-optimum==1.24.0
+optimum==2.0.0
 torch==2.9.0
 torchvision==0.24.0
diff --git a/tests/openvino/requirements.txt b/tests/openvino/requirements.txt
@@ -9,10 +9,12 @@ pytest-xdist
 torch
 torchvision
 virtualenv
+
 addict>=2.4.0
 timm==0.9.2
 efficientnet_pytorch==0.7.1
 datasets==3.0.1
-transformers==4.52.1
-optimum-intel==1.24.0
-optimum==1.26.0
+transformers==4.53.0
+optimum-intel==1.26.0
+optimum-onnx==0.0.3
+optimum==2.0.0
diff --git a/tests/post_training/requirements.txt b/tests/post_training/requirements.txt
@@ -10,14 +10,15 @@ pytest-split
 
 librosa==0.10.0
 memory-profiler==0.61.0
-optimum-intel==1.24.0
-optimum==1.26.0
+optimum-intel==1.26.0
+optimum-onnx==0.0.3
+optimum==2.0.0
 scikit-learn>=1.2.2,<=1.5.0
 soundfile==0.12.1
 tensorboard==2.13.0
 tensorflow-io==0.32.0
 timm==0.9.2
 accelerate==1.9.0
-transformers==4.52.1
+transformers==4.53.0
 whowhatbench @ git+https://github.com/openvinotoolkit/openvino.genai@2025.3.0.0#subdirectory=tools/who_what_benchmark
 datasets==3.6.0
diff --git a/tests/post_training/test_quantize_conformance.py b/tests/post_training/test_quantize_conformance.py
@@ -15,13 +15,11 @@
 import traceback
 from collections import OrderedDict
 from collections import defaultdict
-from functools import partial
 from pathlib import Path
 from typing import Optional
 
 import pandas as pd
 import pytest
-import torch
 import yaml
 from packaging import version
 
@@ -38,8 +36,10 @@
 DATA_ROOT = Path(__file__).parent / "data"
 
 
-# TODO(AlexanderDokuchaev): Remove it after update optimum
-torch.onnx.export = partial(torch.onnx.export, dynamo=False)
+# TODO(AlexanderDokuchaev): WA for https://github.com/huggingface/optimum-intel/issues/1498
+from optimum.exporters.tasks import TasksManager  # noqa: E402
+
+TasksManager._TRANSFORMERS_TASKS_TO_MODEL_LOADERS["image-text-to-text"] = "AutoModelForImageTextToText"
 
 
 @pytest.fixture(scope="function", name="use_avx2")
diff --git a/tests/torch/data/reference_graphs/pruning_groups/Wave2Vec_2.0.dot b/tests/torch/data/reference_graphs/pruning_groups/Wave2Vec_2.0.dot
@@ -1,49 +1,49 @@
-strict digraph  {
-0 [color=red, label="Block: S:1__O:0
+strict digraph {
+0 [label="Block: S:1__O:0
 Producers: 19
-Consumers: ", style=filled];
-1 [color=grey, label="Block: S:1__O:0
+Consumers: ", color=red, style=filled];
+1 [label="Block: S:1__O:0
 Producers: 29
-Consumers: ", style=filled];
-2 [color=red, label="Block: S:24__O:0
+Consumers: ", color=grey, style=filled];
+2 [label="Block: S:24__O:0
 Producers: 29
-Consumers: ", style=filled];
-3 [color=red, label="Block: S:1__O:24
+Consumers: ", color=red, style=filled];
+3 [label="Block: S:1__O:24
 Producers: 29
-Consumers: ", style=filled];
-4 [color=grey, label="Block: S:1__O:0
-Producers: 30
-Consumers: ", style=filled];
-5 [color=red, label="Block: S:24__O:0
-Producers: 30
-Consumers: ", style=filled];
-6 [color=red, label="Block: S:1__O:24
-Producers: 30
-Consumers: ", style=filled];
-7 [color=grey, label="Block: S:1__O:0
-Producers: 34
-Consumers: ", style=filled];
-8 [color=red, label="Block: S:24__O:0
-Producers: 34
-Consumers: ", style=filled];
-9 [color=red, label="Block: S:1__O:24
-Producers: 34
-Consumers: ", style=filled];
-10 [color=red, label="Block: S:1__O:0
-Producers: 44
-Consumers: ", style=filled];
-11 [color=green, label="Block: S:1__O:0
-Producers: 48
-Consumers: 51", style=filled];
-12 [color=red, label="Block: S:1__O:0
-Producers: 51
-Consumers: ", style=filled];
-13 [color=red, label="Block: S:1__O:0
-Producers: 55
-Consumers: ", style=filled];
-14 [color=red, label="Block: S:1__O:0
+Consumers: ", color=red, style=filled];
+4 [label="Block: S:1__O:0
+Producers: 32
+Consumers: ", color=grey, style=filled];
+5 [label="Block: S:24__O:0
+Producers: 32
+Consumers: ", color=red, style=filled];
+6 [label="Block: S:1__O:24
+Producers: 32
+Consumers: ", color=red, style=filled];
+7 [label="Block: S:1__O:0
+Producers: 35
+Consumers: ", color=grey, style=filled];
+8 [label="Block: S:24__O:0
+Producers: 35
+Consumers: ", color=red, style=filled];
+9 [label="Block: S:1__O:24
+Producers: 35
+Consumers: ", color=red, style=filled];
+10 [label="Block: S:1__O:0
+Producers: 46
+Consumers: ", color=red, style=filled];
+11 [label="Block: S:1__O:0
+Producers: 50
+Consumers: 53", color=green, style=filled];
+12 [label="Block: S:1__O:0
+Producers: 53
+Consumers: ", color=red, style=filled];
+13 [label="Block: S:1__O:0
 Producers: 57
-Consumers: ", style=filled];
+Consumers: ", color=red, style=filled];
+14 [label="Block: S:1__O:0
+Producers: 59
+Consumers: ", color=red, style=filled];
 1 -> 2;
 1 -> 3;
 4 -> 5;
diff --git a/tests/torch/data/search_building_block/wave2vec_2.0.json b/tests/torch/data/search_building_block/wave2vec_2.0.json
diff --git a/tests/torch/pruning/experimental/test_nodes_grouping.py b/tests/torch/pruning/experimental/test_nodes_grouping.py
diff --git a/tests/torch/requirements.txt b/tests/torch/requirements.txt
diff --git a/tests/torch/sparsity/movement/test_model_saving.py b/tests/torch/sparsity/movement/test_model_saving.py
diff --git a/tests/torch2/data/function_hook/sparsify_activations/dummy_llama_int8_sym_weights_sparse_activations.dot b/tests/torch2/data/function_hook/sparsify_activations/dummy_llama_int8_sym_weights_sparse_activations.dot
diff --git a/tests/torch2/data/function_hook/sparsify_activations/dummy_llama_sparse_activations.dot b/tests/torch2/data/function_hook/sparsify_activations/dummy_llama_sparse_activations.dot
diff --git a/tests/torch2/requirements.txt b/tests/torch2/requirements.txt
