Merge branch 'main' into release/1.7

Author: Jintao-Huang

README.md

@@ -212,7 +212,7 @@ You can refer to the following scripts to customize your own training script.
 - full: [qwen1half-7b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen1half_7b_chat/full) (A100), [qwen-7b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_7b_chat/full_mp) (2\*A100)
 - full+ddp+zero2: [qwen-7b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_7b_chat/full_ddp_zero2) (4\*A100)
 - full+ddp+zero3: [qwen-14b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_14b_chat/full_ddp_zero3) (4\*A100)
-- lora: [chatglm3-6b](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/chatglm3_6b/lora) (3090), [yi-34b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/yi_34b_chat/lora) (A100), [qwen-72b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_72b_chat/lora_mp) (2\*A100)
+- lora: [chatglm3-6b](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/chatglm3_6b/lora) (3090), [baichuan2-13b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/baichuan2_13b_chat/lora_mp) (2\*3090), [yi-34b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/yi_34b_chat/lora) (A100), [qwen-72b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_72b_chat/lora_mp) (2\*A100)
 - lora+ddp: [chatglm3-6b](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/chatglm3_6b/lora_ddp) (2\*3090)
 - lora+ddp+zero3: [qwen-14b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_14b_chat/lora_ddp_zero3) (4\*3090), [qwen-72b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_72b_chat/lora_ddp_zero3) (4\*A100)
 - qlora(gptq-int4): [qwen-7b-chat-int4](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_7b_chat_int4/qlora) (3090)

README_CN.md

@@ -212,7 +212,7 @@ app_ui_main(infer_args)
 - full: [qwen1half-7b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen1half_7b_chat/full) (A100), [qwen-7b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_7b_chat/full_mp) (2\*A100)
 - full+ddp+zero2: [qwen-7b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_7b_chat/full_ddp_zero2) (4\*A100)
 - full+ddp+zero3: [qwen-14b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_14b_chat/full_ddp_zero3) (4\*A100)
-- lora: [chatglm3-6b](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/chatglm3_6b/lora) (3090), [yi-34b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/yi_34b_chat/lora) (A100), [qwen-72b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_72b_chat/lora_mp) (2\*A100)
+- lora: [chatglm3-6b](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/chatglm3_6b/lora) (3090), [baichuan2-13b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/baichuan2_13b_chat/lora_mp) (2\*3090), [yi-34b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/yi_34b_chat/lora) (A100), [qwen-72b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_72b_chat/lora_mp) (2\*A100)
 - lora+ddp: [chatglm3-6b](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/chatglm3_6b/lora_ddp) (2\*3090)
 - lora+ddp+zero3: [qwen-14b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_14b_chat/lora_ddp_zero3) (4\*3090), [qwen-72b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_72b_chat/lora_ddp_zero3) (4\*A100)
 - qlora(gptq-int4): [qwen-7b-chat-int4](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_7b_chat_int4/qlora) (3090)

docs/source/LLM/LLM微调文档.md

@@ -157,7 +157,7 @@ cd examples/pytorch/llm
 - full: [qwen1half-7b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen1half_7b_chat/full) (A100), [qwen-7b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_7b_chat/full_mp) (2\*A100)
 - full+ddp+zero2: [qwen-7b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_7b_chat/full_ddp_zero2) (4\*A100)
 - full+ddp+zero3: [qwen-14b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_14b_chat/full_ddp_zero3) (4\*A100)
-- lora: [chatglm3-6b](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/chatglm3_6b/lora) (3090), [yi-34b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/yi_34b_chat/lora) (A100), [qwen-72b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_72b_chat/lora_mp) (2\*A100)
+- lora: [chatglm3-6b](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/chatglm3_6b/lora) (3090), [baichuan2-13b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/baichuan2_13b_chat/lora_mp) (2\*3090), [yi-34b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/yi_34b_chat/lora) (A100), [qwen-72b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_72b_chat/lora_mp) (2\*A100)
 - lora+ddp: [chatglm3-6b](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/chatglm3_6b/lora_ddp) (2\*3090)
 - lora+ddp+zero3: [qwen-14b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_14b_chat/lora_ddp_zero3) (4\*3090), [qwen-72b-chat](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_72b_chat/lora_ddp_zero3) (4\*A100)
 - qlora(gptq-int4): [qwen-7b-chat-int4](https://github.com/modelscope/swift/tree/main/examples/pytorch/llm/scripts/qwen_7b_chat_int4/qlora) (3090)

docs/source/LLM/index.md

@@ -10,8 +10,9 @@
 1. [LLM推理文档](LLM推理文档.md)
 2. [LLM微调文档](LLM微调文档.md)
 3. [DPO训练文档](LLM人类对齐训练文档.md)
-4. [LLM量化文档](LLM量化文档.md)
-5. [VLLM推理加速与部署](VLLM推理加速与部署.md)
+4. [界面训练与推理](https://github.com/modelscope/swift/blob/main/docs/source/GetStarted/%E7%95%8C%E9%9D%A2%E8%AE%AD%E7%BB%83%E6%8E%A8%E7%90%86.md)
+5. [LLM量化文档](LLM量化文档.md)
+6. [VLLM推理加速与部署](VLLM推理加速与部署.md)
 
 ### 参考文档
 1. [自定义模型和数据集](自定义与拓展.md)

examples/pytorch/llm/scripts/baichuan2_13b_chat/lora_mp/infer.sh

@@ -0,0 +1,13 @@
+# Experimental environment: 2 * 3090
+
+CUDA_VISIBLE_DEVICES=0,1 \
+swift infer \
+    --ckpt_dir "output/baichuan2-13b-chat/vx-xxx/checkpoint-xxx" \
+    --load_dataset_config true \
+    --max_length 2048 \
+    --max_new_tokens 2048 \
+    --temperature 0.5 \
+    --top_p 0.7 \
+    --repetition_penalty 1. \
+    --do_sample true \
+    --merge_lora false \

examples/pytorch/llm/scripts/baichuan2_13b_chat/lora_mp/sft.sh

@@ -0,0 +1,32 @@
+# Experimental environment: 2 * 3090
+# 2 * 22GB GPU memory
+
+CUDA_VISIBLE_DEVICES=0,1 \
+swift sft \
+    --model_id_or_path baichuan-inc/Baichuan2-13B-Chat \
+    --model_revision master \
+    --sft_type lora \
+    --tuner_backend swift \
+    --template_type baichuan \
+    --dtype AUTO \
+    --output_dir output \
+    --dataset dureader-robust-zh \
+    --train_dataset_sample -1 \
+    --num_train_epochs 5 \
+    --max_length 2048 \
+    --check_dataset_strategy warning \
+    --lora_rank 8 \
+    --lora_alpha 32 \
+    --lora_dropout_p 0.05 \
+    --lora_target_modules W_pack \
+    --gradient_checkpointing true \
+    --batch_size 1 \
+    --weight_decay 0.01 \
+    --learning_rate 1e-4 \
+    --gradient_accumulation_steps 16 \
+    --max_grad_norm 0.5 \
+    --warmup_ratio 0.03 \
+    --eval_steps 100 \
+    --save_steps 100 \
+    --save_total_limit 2 \
+    --logging_steps 10 \

swift/llm/ds_config/zero2.json

@@ -23,8 +23,9 @@
     },
 
     "scheduler": {
-        "type": "WarmupLR",
+        "type": "WarmupDecayLR",
         "params": {
+            "total_num_steps": "auto",
             "warmup_min_lr": "auto",
             "warmup_max_lr": "auto",
             "warmup_num_steps": "auto"

swift/llm/ds_config/zero3.json

@@ -23,8 +23,9 @@
     },
 
     "scheduler": {
-        "type": "WarmupLR",
+        "type": "WarmupDecayLR",
         "params": {
+            "total_num_steps": "auto",
             "warmup_min_lr": "auto",
             "warmup_max_lr": "auto",
             "warmup_num_steps": "auto"

swift/llm/utils/argument.py

@@ -655,7 +655,7 @@ def __post_init__(self) -> None:
             if self.sft_type == 'lora':
                 assert self.merge_lora is True, (
                     'To use VLLM, you need to provide the complete weight parameters. '
-                    'Please set --merge_lora true.')
+                    'Please set `--merge_lora true`.')
         template_info = TEMPLATE_MAPPING[self.template_type]
         support_stream = template_info.get('support_stream', True)
         if self.num_beams != 1 or not support_stream:

tests/llm/data/alpaca.jsonl

@@ -1,3 +1,3 @@
 {"instruction": "11111", "input": "22222", "output": "33333"}
-{"instruction": "aaaaa", "input": "bbbbb", "output": "ccccc"}
+{"instruction": "aaaaa", "output": "ccccc"}
 {"instruction": "AAAAA", "input": "BBBBB", "output": "CCCCC"}