chore: import upstream snapshot with attribution

examples/train/padding_free/dpo_vlm.sh

@@ -0,0 +1,35 @@
+# with padding_free: 4 * 53GiB, 3.55s/it
+# without padding_free: 4 * 62GiB 4.41s/it
+PYTORCH_CUDA_ALLOC_CONF='expandable_segments:True' \
+CUDA_VISIBLE_DEVICES=0,1,2,3 \
+NPROC_PER_NODE=4 \
+MAX_PIXELS=1003520 \
+swift rlhf \
+    --rlhf_type dpo \
+    --model Qwen/Qwen2.5-VL-7B-Instruct \
+    --dataset 'swift/RLAIF-V-Dataset#20000' \
+    --load_from_cache_file true \
+    --split_dataset_ratio 0.01 \
+    --tuner_type full \
+    --torch_dtype bfloat16 \
+    --num_train_epochs 1 \
+    --per_device_train_batch_size 4 \
+    --per_device_eval_batch_size 4 \
+    --learning_rate 1e-5 \
+    --freeze_vit true \
+    --freeze_aligner true \
+    --gradient_accumulation_steps 1 \
+    --eval_steps 100 \
+    --save_steps 100 \
+    --save_total_limit 2 \
+    --deepspeed zero3 \
+    --logging_steps 5 \
+    --max_length 4096 \
+    --output_dir output \
+    --warmup_ratio 0.05 \
+    --dataloader_num_workers 4 \
+    --dataset_num_proc 4 \
+    --attn_impl flash_attn \
+    --save_only_model true \
+    --rpo_alpha 0.1 \
+    --padding_free true

examples/train/padding_free/sft.sh

@@ -0,0 +1,31 @@
+# Supported multimodal models reference:
+# https://github.com/modelscope/ms-swift/blob/main/examples/train/packing/qwen2_5_vl.sh
+# without padding_free: 4 * 60GiB, 26h
+# padding_free: 4 * 44GiB, 13h
+NPROC_PER_NODE=4 \
+CUDA_VISIBLE_DEVICES=0,1,2,3 \
+swift sft \
+    --model Qwen/Qwen2.5-7B \
+    --tuner_type full \
+    --dataset 'liucong/Chinese-DeepSeek-R1-Distill-data-110k-SFT' \
+    --load_from_cache_file true \
+    --split_dataset_ratio 0.01 \
+    --torch_dtype bfloat16 \
+    --per_device_train_batch_size 8 \
+    --per_device_eval_batch_size 8 \
+    --learning_rate 1e-5 \
+    --gradient_accumulation_steps 1 \
+    --eval_steps 200 \
+    --save_steps 200 \
+    --logging_steps 5 \
+    --max_length 8192 \
+    --warmup_ratio 0.05 \
+    --dataloader_num_workers 8 \
+    --dataset_num_proc 8 \
+    --save_total_limit 2 \
+    --save_only_model true \
+    --output_dir output/Qwen2.5-7B \
+    --deepspeed zero3 \
+    --use_liger_kernel true \
+    --attn_impl flash_attn \
+    --padding_free true