chore: import upstream snapshot with attribution

examples/export/quantize/awq.sh

@@ -0,0 +1,14 @@
+pip install "transformers<4.52"
+
+CUDA_VISIBLE_DEVICES=0 \
+swift export \
+    --model Qwen/Qwen2.5-72B-Instruct \
+    --dataset 'AI-ModelScope/alpaca-gpt4-data-zh#500' \
+              'AI-ModelScope/alpaca-gpt4-data-en#500' \
+    --device_map cpu \
+    --quant_n_samples 256 \
+    --quant_batch_size 1 \
+    --max_length 2048 \
+    --quant_method awq \
+    --quant_bits 4 \
+    --output_dir Qwen2.5-72B-Instruct-AWQ

examples/export/quantize/bert/bnb.sh

@@ -0,0 +1,16 @@
+# merge-lora
+CUDA_VISIBLE_DEVICES=0 swift export \
+    --adapters swift/test_bert \
+    --output_dir output/swift_test_bert_merged \
+    --merge_lora true
+
+# bnb quantize
+CUDA_VISIBLE_DEVICES=0 swift export \
+    --model output/swift_test_bert_merged \
+    --output_dir output/swift_test_bert_bnb_int4 \
+    --quant_bits 4 \
+    --quant_method bnb
+
+# infer
+CUDA_VISIBLE_DEVICES=0 swift infer \
+    --model output/swift_test_bert_bnb_int4

examples/export/quantize/bert/gptq.sh

@@ -0,0 +1,33 @@
+# merge-lora
+CUDA_VISIBLE_DEVICES=0 swift export \
+    --adapters swift/test_bert \
+    --output_dir output/swift_test_bert_merged \
+    --merge_lora true
+
+EXIT_CODE=$?
+
+if [ $EXIT_CODE -ne 0 ]; then
+    echo "Error: LoRA merge failed with exit code $EXIT_CODE"
+    exit $EXIT_CODE
+fi
+
+# gptq quantize
+CUDA_VISIBLE_DEVICES=0 swift export \
+    --model output/swift_test_bert_merged \
+    --load_data_args true \
+    --output_dir output/swift_test_bert_gptq_int4 \
+    --quant_bits 4 \
+    --quant_method gptq \
+    --max_length 512
+
+
+EXIT_CODE=$?
+
+if [ $EXIT_CODE -ne 0 ]; then
+    echo "Error: GPTQ quantization failed with exit code $EXIT_CODE"
+    exit $EXIT_CODE
+fi
+
+# infer
+CUDA_VISIBLE_DEVICES=0 swift infer \
+    --model output/swift_test_bert_gptq_int4

examples/export/quantize/bnb.sh

@@ -0,0 +1,8 @@
+CUDA_VISIBLE_DEVICES=0 \
+swift export \
+    --model Qwen/Qwen2.5-1.5B-Instruct \
+    --quant_method bnb \
+    --quant_bits 4 \
+    --bnb_4bit_quant_type nf4 \
+    --bnb_4bit_use_double_quant true \
+    --output_dir Qwen2.5-1.5B-Instruct-BNB-NF4

examples/export/quantize/fp8.sh

@@ -0,0 +1,9 @@
+# Due to the structural changes made to MoE architecture in `transformers>=5.0`,
+# if you need to apply FP8 quantization to MoE models, please use `megatron export`
+# (compatible with vLLM inference).
+# Reference: https://github.com/modelscope/ms-swift/blob/main/examples/megatron/fp8/quant.sh
+CUDA_VISIBLE_DEVICES=0 \
+swift export \
+    --model Qwen/Qwen2.5-3B-Instruct \
+    --quant_method fp8 \
+    --output_dir Qwen2.5-3B-Instruct-FP8

examples/export/quantize/gptq.sh

@@ -0,0 +1,13 @@
+# OMP_NUM_THREADS=14 please Check issue: https://github.com/AutoGPTQ/AutoGPTQ/issues/439
+OMP_NUM_THREADS=14 \
+CUDA_VISIBLE_DEVICES=0 \
+swift export \
+    --model Qwen/Qwen2.5-1.5B-Instruct \
+    --dataset 'AI-ModelScope/alpaca-gpt4-data-zh#500' \
+              'AI-ModelScope/alpaca-gpt4-data-en#500' \
+    --quant_n_samples 256 \
+    --quant_batch_size 1 \
+    --max_length 2048 \
+    --quant_method gptq \
+    --quant_bits 4 \
+    --output_dir Qwen2.5-1.5B-Instruct-GPTQ-Int4

examples/export/quantize/gptq_v2.sh

@@ -0,0 +1,14 @@
+# You need to install gptqmodel.
+# OMP_NUM_THREADS=14 please Check issue: https://github.com/AutoGPTQ/AutoGPTQ/issues/439
+OMP_NUM_THREADS=14 \
+CUDA_VISIBLE_DEVICES=0 \
+swift export \
+    --model Qwen/Qwen2.5-1.5B-Instruct \
+    --dataset 'AI-ModelScope/alpaca-gpt4-data-zh#500' \
+              'AI-ModelScope/alpaca-gpt4-data-en#500' \
+    --quant_n_samples 256 \
+    --quant_batch_size 1 \
+    --max_length 2048 \
+    --quant_method gptq_v2 \
+    --quant_bits 4 \
+    --output_dir Qwen2.5-1.5B-Instruct-GPTQ-V2-Int4

examples/export/quantize/mllm/awq.sh

@@ -0,0 +1,18 @@
+pip install "transformers==4.51.*"
+
+CUDA_VISIBLE_DEVICES=0 \
+MAX_PIXELS=1003520 \
+VIDEO_MAX_PIXELS=50176 \
+FPS_MAX_FRAMES=12 \
+swift export \
+    --model Qwen/Qwen2.5-VL-3B-Instruct \
+    --dataset 'AI-ModelScope/alpaca-gpt4-data-zh#500' \
+              'AI-ModelScope/alpaca-gpt4-data-en#500' \
+              'modelscope/coco_2014_caption:validation#500' \
+              'swift/VideoChatGPT:Generic#500' \
+    --quant_n_samples 256 \
+    --quant_batch_size -1 \
+    --max_length 2048 \
+    --quant_method awq \
+    --quant_bits 4 \
+    --output_dir Qwen2.5-VL-3B-Instruct-AWQ

examples/export/quantize/mllm/bnb.sh

@@ -0,0 +1,6 @@
+CUDA_VISIBLE_DEVICES=0 \
+swift export \
+    --model Qwen/Qwen2.5-VL-3B-Instruct \
+    --quant_method bnb \
+    --quant_bits 4 \
+    --output_dir Qwen2.5-VL-3B-Instruct-BNB-Int4

examples/export/quantize/mllm/fp8.sh

@@ -0,0 +1,10 @@
+# use transformers==5.2.0
+CUDA_VISIBLE_DEVICES=0 \
+swift export \
+    --model Qwen/Qwen3.5-4B \
+    --quant_method fp8 \
+    --output_dir Qwen3.5-4B-FP8
+
+# CUDA_VISIBLE_DEVICES=0 \
+# swift infer \
+#     --model Qwen3.5-4B-FP8

examples/export/quantize/mllm/gptq.sh

@@ -0,0 +1,18 @@
+# OMP_NUM_THREADS=14 please Check issue: https://github.com/AutoGPTQ/AutoGPTQ/issues/439
+OMP_NUM_THREADS=14 \
+CUDA_VISIBLE_DEVICES=0 \
+MAX_PIXELS=1003520 \
+VIDEO_MAX_PIXELS=50176 \
+FPS_MAX_FRAMES=12 \
+swift export \
+    --model Qwen/Qwen2.5-VL-3B-Instruct \
+    --dataset 'AI-ModelScope/alpaca-gpt4-data-zh#500' \
+              'AI-ModelScope/alpaca-gpt4-data-en#500' \
+              'modelscope/coco_2014_caption:validation#500' \
+              'swift/VideoChatGPT:Generic#500' \
+    --quant_n_samples 256 \
+    --quant_batch_size 1 \
+    --max_length 2048 \
+    --quant_method gptq \
+    --quant_bits 4 \
+    --output_dir Qwen2.5-VL-3B-Instruct-GPTQ-Int4

examples/export/quantize/moe/awq.sh

@@ -0,0 +1,13 @@
+pip install "transformers<4.52"
+
+CUDA_VISIBLE_DEVICES=0,1 \
+swift export \
+    --model Qwen/Qwen3-30B-A3B \
+    --dataset 'swift/Qwen3-SFT-Mixin' \
+    --device_map auto \
+    --quant_n_samples 64 \
+    --quant_batch_size -1 \
+    --max_length 8192 \
+    --quant_method awq \
+    --quant_bits 4 \
+    --output_dir Qwen3-30B-A3B-AWQ

examples/export/quantize/moe/bnb.sh

@@ -0,0 +1,6 @@
+CUDA_VISIBLE_DEVICES=0 \
+swift export \
+    --model Qwen/Qwen3-30B-A3B \
+    --quant_method bnb \
+    --quant_bits 4 \
+    --output_dir Qwen3-30B-A3B-BNB-Int4

examples/export/quantize/moe/fp8.sh

@@ -0,0 +1,11 @@
+CUDA_VISIBLE_DEVICES=0 \
+swift export \
+    --model Qwen/Qwen3-30B-A3B \
+    --quant_method fp8 \
+    --output_dir Qwen3-30B-A3B-FP8
+
+# CUDA_VISIBLE_DEVICES=0 \
+# swift infer \
+#     --model Qwen3-30B-A3B-FP8 \
+#     --infer_backend vllm \
+#     --stream true

examples/export/quantize/moe/gptq.sh

@@ -0,0 +1,13 @@
+# 2 * 80GB
+OMP_NUM_THREADS=14 \
+CUDA_VISIBLE_DEVICES=0,1 \
+swift export \
+    --model Qwen/Qwen2-57B-A14B-Instruct \
+    --dataset 'AI-ModelScope/alpaca-gpt4-data-zh#1000' \
+              'AI-ModelScope/alpaca-gpt4-data-en#1000' \
+    --quant_n_samples 512 \
+    --quant_batch_size 1 \
+    --max_length 4096 \
+    --quant_method gptq \
+    --quant_bits 4 \
+    --output_dir Qwen2-57B-A14B-Instruct-GPTQ-Int4

examples/export/quantize/omni/gptq.sh

@@ -0,0 +1,18 @@
+# OMP_NUM_THREADS=14 please Check issue: https://github.com/AutoGPTQ/AutoGPTQ/issues/439
+OMP_NUM_THREADS=14 \
+CUDA_VISIBLE_DEVICES=0 \
+MAX_PIXELS=1003520 \
+VIDEO_MAX_PIXELS=50176 \
+FPS_MAX_FRAMES=12 \
+swift export \
+    --model Qwen/Qwen2.5-Omni-7B \
+    --dataset 'AI-ModelScope/alpaca-gpt4-data-zh#500' \
+              'AI-ModelScope/alpaca-gpt4-data-en#500' \
+              'modelscope/coco_2014_caption:validation#500' \
+              'swift/VideoChatGPT:Generic#500' \
+    --quant_n_samples 256 \
+    --quant_batch_size 1 \
+    --max_length 2048 \
+    --quant_method gptq \
+    --quant_bits 4 \
+    --output_dir Qwen2.5-Omni-7B-GPTQ-Int4

examples/export/quantize/reward_model/bnb.sh

@@ -0,0 +1,12 @@
+# bnb quantize
+CUDA_VISIBLE_DEVICES=0 swift export \
+    --model Shanghai_AI_Laboratory/internlm2-1_8b-reward \
+    --output_dir output/internlm2-1_8b-reward-bnb-int4 \
+    --quant_bits 4 \
+    --quant_method bnb
+
+# infer
+CUDA_VISIBLE_DEVICES=0 swift infer \
+    --model output/internlm2-1_8b-reward-bnb-int4 \
+    --val_dataset 'AI-ModelScope/alpaca-gpt4-data-zh#1000' \
+    --max_batch_size 16

examples/export/quantize/reward_model/gptq.sh

@@ -0,0 +1,14 @@
+# gptq quantize
+CUDA_VISIBLE_DEVICES=0 swift export \
+    --model Shanghai_AI_Laboratory/internlm2-1_8b-reward \
+    --output_dir output/internlm2-1_8b-reward-gptq-int4 \
+    --quant_bits 4 \
+    --max_length 2048 \
+    --quant_method gptq \
+    --dataset 'AI-ModelScope/alpaca-gpt4-data-zh#1000' 'AI-ModelScope/alpaca-gpt4-data-en#1000'
+
+# infer
+CUDA_VISIBLE_DEVICES=0 swift infer \
+    --model output/internlm2-1_8b-reward-gptq-int4 \
+    --val_dataset 'AI-ModelScope/alpaca-gpt4-data-zh#1000' \
+    --max_batch_size 16