Refactor code structure for improved readability and maintainability

dataset/OCR-VQA-200K/download.py

@@ -6,24 +6,25 @@ import concurrent.futures
 import threading
 
 # Set the file paths for your Google Drive
-dataset_path = './dataset.json'
+dataset_path = "./dataset.json"
-images_path = './images'
+images_path = "./images"
 download = 1  # Set to 0 if images are already downloaded
 
 # Load dataset json file
-with open(dataset_path, 'r') as fp:
+with open(dataset_path, "r") as fp:
     data = json.load(fp)
 
 # Initialize a counter and a lock for thread-safe counting
 downloaded_count = 0
 count_lock = threading.Lock()
 
+
 # Function to download an image
 def download_image(k):
     global downloaded_count
-    imageURL = data[k]['imageURL']
+    imageURL = data[k]["imageURL"]
     ext = os.path.splitext(imageURL)[1]
-    outputFile = os.path.join(images_path, f'{k}{ext}')
+    outputFile = os.path.join(images_path, f"{k}{ext}")
 
     # Only download the image if it doesn't exist
     if not os.path.exists(outputFile):
@@ -33,9 +34,10 @@ def download_image(k):
             with count_lock:
                 downloaded_count += 1
                 if downloaded_count % 100 == 0:
-                    print(f'{downloaded_count} images downloaded.')
+                    print(f"{downloaded_count} images downloaded.")
         except urllib.error.URLError as e:
-            print(f'Error downloading {outputFile}: {e}')
+            print(f"Error downloading {outputFile}: {e}")
+
 
 # Download images using multiple threads
 if download == 1:
@@ -45,5 +47,5 @@ if download == 1:
     # Create a thread pool and download the images in parallel
     # Increase max_workers to potentially speed up downloads for many small files.
     # The optimal number may vary based on your network and the server's capacity.
-    with concurrent.futures.ThreadPoolExecutor(max_workers=50) as executor:
+    with concurrent.futures.ThreadPoolExecutor(max_workers=400) as executor:
         executor.map(download_image, data.keys())

pyproject.toml

@@ -5,6 +5,7 @@ dependencies = [
   "datasets==3.2.0",
   "deepspeed==0.16.2",
   "evaluate==0.4.3",
+  "huggingface-hub==0.30.1",
   "librosa>=0.10.2.post1",
   "markupsafe==2.1.5",
   "ms-swift>=1.3.0",

src/.gitignore

@@ -1 +1,2 @@
-checkpoint/*
+checkpoint/*
+wandb/*

src/configs/accelerate_configs/deepspeed_zero1.yaml

@@ -2,7 +2,7 @@ compute_environment: LOCAL_MACHINE
 debug: false
 deepspeed_config:
   deepspeed_multinode_launcher: standard
-  gradient_accumulation_steps: 1
+  gradient_accumulation_steps: 4
   zero3_init_flag: false
   zero_stage: 1
 distributed_type: DEEPSPEED
@@ -11,7 +11,7 @@ machine_rank: 0
 main_training_function: main
 mixed_precision: 'bf16'
 num_machines: 1
-num_processes: 8
+num_processes: 4
 rdzv_backend: static
 same_network: true
 tpu_env: []

src/configs/accelerate_configs/deepspeed_zero2.yaml

@@ -12,7 +12,7 @@ machine_rank: 0
 main_training_function: main
 mixed_precision: 'bf16'
 num_machines: 1
-num_processes: 8
+num_processes: 4
 rdzv_backend: static
 same_network: true
 tpu_env: []

src/evaluation.py

@@ -38,12 +38,44 @@ if __name__ == "__main__":
 
     from model_library.factory import get_model
 
-    if model_args.model_name_or_path == "Qwen/Qwen2-VL-7B-Instruct":
+    if model_args.model_name_or_path == "Qwen/Qwen2.5-VL-3B-Instruct":
         torch_dtype = (
             model_args.torch_dtype
             if model_args.torch_dtype in ["auto", None]
             else getattr(torch, model_args.torch_dtype)
         )
+        quantization_config = get_quantization_config(model_args)
+        model_kwargs = dict(
+            attn_implementation=model_args.attn_implementation,
+            torch_dtype=torch_dtype,
+            quantization_config=quantization_config,
+        )
+        from transformers import Qwen2_5_VLProcessor, Qwen2_5_VLForConditionalGeneration
+
+        model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
+            training_args.output_dir,
+            **model_kwargs,
+        )
+
+        processor = Qwen2_5_VLProcessor.from_pretrained(
+            model_args.model_name_or_path,
+            trust_remote_code=model_args.trust_remote_code,
+            padding_side="left",
+        )
+
+        from model_library.qwen2vl import collate_fn_for_train, collate_fn_for_evaluate
+        from functools import partial
+
+        collate_fn_for_train = partial(collate_fn_for_train, processor=processor)
+        collate_fn_for_evaluate = partial(collate_fn_for_evaluate, processor=processor)
+
+    elif model_args.model_name_or_path == "Qwen/Qwen2-VL-7B-Instruct":
+        torch_dtype = (
+            model_args.torch_dtype
+            if model_args.torch_dtype in ["auto", None]
+            else getattr(torch, model_args.torch_dtype)
+        )
+
         quantization_config = get_quantization_config(model_args)
         model_kwargs = dict(
             attn_implementation=model_args.attn_implementation,

src/model_library/factory.py

@@ -68,4 +68,25 @@ def get_model(model_args: ContinualModelConfig):
         collate_fn_for_train = partial(collate_fn_for_train, processor=processor)
         collate_fn_for_evaluate = partial(collate_fn_for_evaluate, processor=processor)
 
+    if model_args.model_name_or_path == "Qwen/Qwen2.5-VL-3B-Instruct":
+        from transformers import Qwen2_5_VLProcessor, Qwen2_5_VLForConditionalGeneration
+
+        model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
+            model_args.model_name_or_path,
+            trust_remote_code=model_args.trust_remote_code,
+            **model_kwargs,
+        )
+
+        processor = Qwen2_5_VLProcessor.from_pretrained(
+            model_args.model_name_or_path,
+            trust_remote_code=model_args.trust_remote_code,
+            padding_side="left",
+        )
+
+        from model_library.qwen2vl import collate_fn_for_train, collate_fn_for_evaluate
+        from functools import partial
+
+        collate_fn_for_train = partial(collate_fn_for_train, processor=processor)
+        collate_fn_for_evaluate = partial(collate_fn_for_evaluate, processor=processor)
+
     return model, processor, collate_fn_for_train, collate_fn_for_evaluate

src/model_library/qwen2vl/__init__.py

@@ -1,5 +1,6 @@
 from .collate_fn import collate_fn_for_evaluate, collate_fn_for_train
-from .model import Qwen2VLForConditionalGeneration_modified
+
+# from .model import Qwen2VLForConditionalGeneration_modified
 
 __all__ = [
     "collate_fn_for_train",

src/peft_library/mapping.py

@@ -73,7 +73,6 @@ from peft.tuners import (
 from .tuners import MMOELoraModel, MMOELoraConfig
 from peft.tuners.tuners_utils import BaseTuner
 from peft.utils import _prepare_prompt_learning_config
-from peft.utils.constants import PEFT_TYPE_TO_PREFIX_MAPPING
 
 
 if TYPE_CHECKING:

src/peft_library/peft_model.py

@@ -46,7 +46,7 @@ from transformers.modeling_outputs import (
 )
 from transformers.utils import PushToHubMixin
 
-from peft.utils.constants import DUMMY_MODEL_CONFIG, PEFT_TYPE_TO_PREFIX_MAPPING
+from peft.utils.constants import DUMMY_MODEL_CONFIG
 
 from peft import __version__
 from peft.config import PeftConfig

src/peft_repo

@@ -1 +1 @@
-Subproject commit 65c3c43cd195bd90b8cb339c1ba883b4c6c66b43
+Subproject commit 83111347f3df66f04bd6759b1a3dcce719380628

src/train.sh

@@ -1,15 +1,15 @@
 #!/bin/bash
 
-accelerate launch --config_file configs/accelerate_configs/deepspeed_zero2.yaml train.py \
+accelerate launch --config_file configs/accelerate_configs/deepspeed_zero1.yaml train.py \
-    --dataset_name chem \
+    --dataset_name refcoco \
     --use_peft \
     --peft_type LORA \
-    --model_name_or_path Qwen/Qwen2-VL-7B-Instruct \
+    --model_name_or_path Qwen/Qwen2.5-VL-3B-Instruct \
     --lora_target_modules .\*proj.\*\|.\*fc.\*\|.\*mlp\.0\|.\*mlp\.2 \
     --lora_r 8 \
     --lora_alpha 32 \
-    --per_device_train_batch_size 2 \
+    --per_device_train_batch_size 1 \
-    --per_device_eval_batch_size 2 \
+    --per_device_eval_batch_size 1 \
     --gradient_accumulation_steps 4 \
     --output_dir checkpoint/qwen2_alllinear/ \
     --learning_rate 1e-4 \

src/transformers_repo

@@ -1 +1 @@
-Subproject commit 7961d291b338d568fa2160f7deac85baa21c49dc
+Subproject commit 684f12be1c8f26c46b1eebad50ce21ce6e3378b3

src/utils/trainer.py

@@ -1,15 +1,6 @@
 # _________________________________________________________
 
 
-from transformers.trainer import (
-    Trainer,
-    _is_peft_model,
-    MODEL_FOR_CAUSAL_LM_MAPPING_NAMES,
-    tpu_spmd_dataloader,
-    logger,
-    has_length,
-    sys,
-)
 from transformers.trainer import *
 from transformers import (
     TrainingArguments,
@@ -32,12 +23,12 @@ class ContinualTrainer(Trainer):
         self.accelerator = accelerator
         super().__init__(model, args, data_collator, train_dataset, eval_dataset)
 
-        if regularization_args.ewc_enable:
+        # if regularization_args.ewc_enable:
-            self.ewc_lambda = regularization_args.ewc_lambda
+        #     self.ewc_lambda = regularization_args.ewc_lambda
-            # fisher = t
+        #     # fisher = t
 
-        if regularization_args.lwf_enable:
+        # if regularization_args.lwf_enable:
-            self.lwf_lambda = regularization_args.lwf_lambda
+        #     self.lwf_lambda = regularization_args.lwf_lambda
 
     def create_accelerator_and_postprocess(self):