chore: import upstream snapshot with attribution

examples/train/agent/deepseek_r1.sh

@@ -0,0 +1,29 @@
+CUDA_VISIBLE_DEVICES=0 \
+swift sft \
+    --model deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B \
+    --tuner_type full \
+    --dataset AI-ModelScope/function-calling-chatml \
+    --load_from_cache_file true \
+    --split_dataset_ratio 0.01 \
+    --agent_template react_en \
+    --loss_scale react \
+    --response_prefix '' \
+    --torch_dtype bfloat16 \
+    --num_train_epochs 2 \
+    --per_device_train_batch_size 1 \
+    --per_device_eval_batch_size 1 \
+    --learning_rate 1e-5 \
+    --gradient_accumulation_steps 8 \
+    --eval_steps 100 \
+    --save_steps 100 \
+    --save_total_limit 2 \
+    --logging_steps 5 \
+    --max_length 8192 \
+    --save_only_model true \
+    --packing true \
+    --use_liger_kernel true \
+    --output_dir output \
+    --warmup_ratio 0.05 \
+    --attn_impl flash_attn \
+    --dataloader_num_workers 4 \
+    --dataset_num_proc 16

examples/train/agent/glm4.sh

@@ -0,0 +1,30 @@
+# 4 * 80GiB
+NPROC_PER_NODE=4 \
+CUDA_VISIBLE_DEVICES=0,1,2,3 \
+swift sft \
+    --model ZhipuAI/GLM-4-9B-0414 \
+    --tuner_type full \
+    --dataset AI-ModelScope/function-calling-chatml \
+    --load_from_cache_file true \
+    --split_dataset_ratio 0.01 \
+    --agent_template hermes \
+    --torch_dtype bfloat16 \
+    --num_train_epochs 2 \
+    --per_device_train_batch_size 1 \
+    --per_device_eval_batch_size 1 \
+    --learning_rate 1e-5 \
+    --gradient_accumulation_steps 2 \
+    --eval_steps 100 \
+    --save_steps 100 \
+    --save_total_limit 2 \
+    --logging_steps 5 \
+    --max_length 8192 \
+    --save_only_model true \
+    --packing true \
+    --deepspeed zero3 \
+    --use_liger_kernel true \
+    --output_dir output \
+    --warmup_ratio 0.05 \
+    --attn_impl flash_attn \
+    --dataloader_num_workers 4 \
+    --dataset_num_proc 16

examples/train/agent/loss_scale/infer_lora.py

@@ -0,0 +1,89 @@
+# Copyright (c) ModelScope Contributors. All rights reserved.
+import os
+
+os.environ['CUDA_VISIBLE_DEVICES'] = '0'
+# os.environ['SWIFT_DEBUG'] = '1'
+
+
+def infer(engine: 'InferEngine', infer_request: 'InferRequest'):
+    stop = [engine.template.agent_template.keyword.observation]  # compat react_en
+    request_config = RequestConfig(max_tokens=512, temperature=0, stop=stop)
+    resp_list = engine.infer([infer_request], request_config)
+    query = infer_request.messages[0]['content']
+    response = resp_list[0].choices[0].message.content
+    print(f'query: {query}')
+    print(f'response: {response}')
+    print(f'tool_calls: {resp_list[0].choices[0].message.tool_calls}')
+
+    tool = '{"temperature": 32, "condition": "Sunny", "humidity": 50}'
+    print(f'tool_response: {tool}')
+    infer_request.messages += [{'role': 'assistant', 'content': response}, {'role': 'tool', 'content': tool}]
+    resp_list = engine.infer([infer_request], request_config)
+    response2 = resp_list[0].choices[0].message.content
+    print(f'response2: {response2}')
+
+
+def infer_stream(engine: 'InferEngine', infer_request: 'InferRequest'):
+    stop = [engine.template.agent_template.keyword.observation]
+    request_config = RequestConfig(max_tokens=512, temperature=0, stream=True, stop=stop)
+    gen_list = engine.infer([infer_request], request_config)
+    query = infer_request.messages[0]['content']
+    response = ''
+    print(f'query: {query}\nresponse: ', end='')
+    for resp in gen_list[0]:
+        if resp is None:
+            continue
+        delta = resp.choices[0].delta.content
+        response += delta
+        print(delta, end='', flush=True)
+    print()
+    print(f'tool_calls: {resp.choices[0].delta.tool_calls}')
+
+    tool = '{"temperature": 32, "condition": "Sunny", "humidity": 50}'
+    print(f'tool_response: {tool}\nresponse2: ', end='')
+    infer_request.messages += [{'role': 'assistant', 'content': response}, {'role': 'tool', 'content': tool}]
+    gen_list = engine.infer([infer_request], request_config)
+    for resp in gen_list[0]:
+        if resp is None:
+            continue
+        print(resp.choices[0].delta.content, end='', flush=True)
+    print()
+
+
+def get_infer_request():
+    return InferRequest(
+        messages=[{
+            'role': 'user',
+            'content': "How's the weather in Beijing today?"
+        }],
+        tools=[{
+            'name': 'get_current_weather',
+            'description': 'Get the current weather in a given location',
+            'parameters': {
+                'type': 'object',
+                'properties': {
+                    'location': {
+                        'type': 'string',
+                        'description': 'The city and state, e.g. San Francisco, CA'
+                    },
+                    'unit': {
+                        'type': 'string',
+                        'enum': ['celsius', 'fahrenheit']
+                    }
+                },
+                'required': ['location']
+            }
+        }])
+
+
+if __name__ == '__main__':
+    from swift.agent_template import agent_template_map
+    from swift.infer_engine import InferEngine, InferRequest, RequestConfig, TransformersEngine
+    model = 'Qwen/Qwen2.5-3B'
+    adapters = ['output/vx-xxx/checkpoint-xxx']
+    engine = TransformersEngine(model, adapters=adapters, max_batch_size=8)
+
+    # engine.template._agent_template = 'hermes'  # react_en/qwen_en/qwen_en_parallel
+
+    infer(engine, get_infer_request())
+    infer_stream(engine, get_infer_request())

examples/train/agent/loss_scale/train.sh

@@ -0,0 +1,30 @@
+# 20GB
+CUDA_VISIBLE_DEVICES=0 \
+swift sft \
+    --model Qwen/Qwen2.5-3B \
+    --tuner_type lora \
+    --dataset AI-ModelScope/function-calling-chatml#10000 \
+    --load_from_cache_file true \
+    --split_dataset_ratio 0.01 \
+    --loss_scale hermes \
+    --agent_template hermes \
+    --torch_dtype bfloat16 \
+    --num_train_epochs 2 \
+    --per_device_train_batch_size 1 \
+    --per_device_eval_batch_size 1 \
+    --learning_rate 1e-4 \
+    --lora_rank 8 \
+    --lora_alpha 32 \
+    --target_modules all-linear \
+    --modules_to_save embed_tokens lm_head \
+    --gradient_accumulation_steps 16 \
+    --eval_steps 100 \
+    --save_steps 100 \
+    --save_total_limit 2 \
+    --logging_steps 5 \
+    --max_length 2048 \
+    --use_liger_kernel true \
+    --output_dir output \
+    --warmup_ratio 0.05 \
+    --dataloader_num_workers 4 \
+    --dataset_num_proc 16

examples/train/agent/qwen2_5.sh

@@ -0,0 +1,28 @@
+# 35GiB
+CUDA_VISIBLE_DEVICES=0 \
+swift sft \
+    --model Qwen/Qwen2.5-3B \
+    --tuner_type full \
+    --dataset AI-ModelScope/function-calling-chatml \
+    --load_from_cache_file true \
+    --split_dataset_ratio 0.01 \
+    --agent_template hermes \
+    --torch_dtype bfloat16 \
+    --num_train_epochs 2 \
+    --per_device_train_batch_size 1 \
+    --per_device_eval_batch_size 1 \
+    --learning_rate 1e-5 \
+    --gradient_accumulation_steps 8 \
+    --eval_steps 100 \
+    --save_steps 100 \
+    --save_total_limit 2 \
+    --logging_steps 5 \
+    --max_length 8192 \
+    --save_only_model true \
+    --packing true \
+    --use_liger_kernel true \
+    --output_dir output \
+    --warmup_ratio 0.05 \
+    --attn_impl flash_attn \
+    --dataloader_num_workers 4 \
+    --dataset_num_proc 16