toolace

.gitignore

+__pycache__

Dockerfile

+FROM image.sourcefind.cn:5000/dcu/admin/base/pytorch:2.4.1-ubuntu22.04-dtk25.04-py3.10
\ No newline at end of file

README.md

+# ToolACE
+
+## 论文
+
+`ToolACE: Winning the Points of LLM Function Calling`
+
+* https://arxiv.org/pdf/2409.00920
+
+
+## 模型结构
+
+模型LLama3.1训练
+
+![alt text](readme_imgs/arch.png)
+
+## 算法原理
+Multi-Head Attention是一种并行注意力机制，它通过多个子空间中的注意力头协同工作，从不同角度捕捉序列中元素之间的关系，从而增强模型的表达能力。
+
+![alt text](readme_imgs/alg.png)
+
+## 环境配置
+
+### Docker（方法一）
+    
+    docker pull image.sourcefind.cn:5000/dcu/admin/base/pytorch:2.4.1-ubuntu22.04-dtk25.04-py3.10
+
+    docker run --shm-size 100g --network=host --name=toolace --privileged --device=/dev/kfd --device=/dev/dri --group-add video --cap-add=SYS_PTRACE --security-opt seccomp=unconfined -v 项目地址(绝对路径):/home/ -v /opt/hyhal:/opt/hyhal:ro -it <your IMAGE ID> bash
+
+    pip install -U transformers
+    pip install accelerate
+
+
+### Dockerfile（方法二）
+
+    docker build -t <IMAGE_NAME>:<TAG> .
+
+    docker run --shm-size 100g --network=host --name=toolace --privileged --device=/dev/kfd --device=/dev/dri --group-add video --cap-add=SYS_PTRACE --security-opt seccomp=unconfined -v 项目地址(绝对路径):/home/ -v /opt/hyhal:/opt/hyhal:ro -it <your IMAGE ID> bash
+    
+    pip install -U transformers
+    pip install accelerate
+
+### Anaconda（方法三）
+
+1、关于本项目DCU显卡所需的特殊深度学习库可从光合开发者社区下载安装： https://developer.sourcefind.cn/tool/
+
+```
+DTK驱动:dtk25.04
+python:python3.10
+torch:2.4.1
+triton:3.0
+flash-attn:2.6.1
+deepspeed:0.14.2
+apex:1.4.0
+```
+
+2、其他非特殊库直接按照requirements.txt安装
+
+```
+pip install -U transformers
+pip install accelerate
+```
+
+## 数据集
+
+[ToolACE](https://huggingface.co/datasets/Team-ACE/ToolACE)
+
+本项目提供适用于`llama-factory`的数据格式，位于`datasets/data_1.json`
+
+
+## 训练
+
+### 安装llama-factory
+
+详情参考[llama-factory](https://developer.sourcefind.cn/codes/OpenDAS/llama-factory/-/tree/0.9.2-parallel_tool).
+
+### Lora
+
+```bash
+llamafactory-cli train examples/train_lora/llama3_lora_sft_ds3.yaml
+```
+
+在运行前需要修改`examples/train_lora/llama3_lora_sft_ds3.yaml`中相应参数。
+
+
+## 推理
+
+```bash
+python inference.py
+```
+
+## result
+
+```python
+[sales_growth.calculate(company="XYZ", years=3), financial_ratios interest_coverage(company_name="XYZ", years=3)]
+```
+
+注意：结果可能根据参数不同而变化。
+
+### 精度
+
+loss: 0.3
+
+## 应用场景
+
+### 算法类别
+
+`对话问答`
+
+### 热点应用行业
+
+`电商,教育,广媒`
+
+## 预训练权重
+
+[ToolACE-2-Llama-3.1-8B](https://hf-mirror.com/Team-ACE/ToolACE-2-Llama-3.1-8B)
+
+## 源码仓库及问题反馈
+
+* 
+
+## 参考资料
+
+* https://hf-mirror.com/Team-ACE/ToolACE-2-Llama-3.1-8B

inference.py

+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+model_name = "Team-ACE/ToolACE-2-Llama-3.1-8B"
+
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(
+    model_name,
+    torch_dtype='auto',
+    device_map='auto'
+)
+
+
+# You can modify the prompt for your task
+system_prompt = """You are an expert in composing functions. You are given a question and a set of possible functions. Based on the question, you will need to make one or more function/tool calls to achieve the purpose.
+If none of the function can be used, point it out. If the given question lacks the parameters required by the function, also point it out.
+You should only return the function call in tools call sections.
+
+If you decide to invoke any of the function(s), you MUST put it in the format of [func_name1(params_name1=params_value1, params_name2=params_value2...), func_name2(params)]
+You SHOULD NOT include any other text in the response.
+Here is a list of functions in JSON format that you can invoke.\n{functions}\n
+"""
+
+# User query
+query = "Find me the sales growth rate for company XYZ for the last 3 years and also the interest coverage ratio for the same duration."
+
+# Availabel tools in JSON format (OpenAI-format)
+tools = [
+    {
+        "name": "financial_ratios.interest_coverage", "description": "Calculate a company's interest coverage ratio given the company name and duration",
+        "arguments": {
+            "type": "dict",
+            "properties": {
+                "company_name": {
+                    "type": "string",
+                    "description": "The name of the company."
+                }, 
+                "years": {
+                    "type": "integer",
+                    "description": "Number of past years to calculate the ratio."
+                }
+            }, 
+            "required": ["company_name", "years"]
+        }
+    },
+    {
+        "name": "sales_growth.calculate",
+        "description": "Calculate a company's sales growth rate given the company name and duration",
+        "arguments": {
+            "type": "dict", 
+            "properties": {
+                "company": {
+                    "type": "string",
+                    "description": "The company that you want to get the sales growth rate for."
+                }, 
+                "years": {
+                    "type": "integer",
+                    "description": "Number of past years for which to calculate the sales growth rate."
+                }
+            }, 
+            "required": ["company", "years"]
+        }
+    },
+    {
+        "name": "weather_forecast",
+        "description": "Retrieve a weather forecast for a specific location and time frame.",
+        "arguments": {
+            "type": "dict",
+            "properties": {
+                "location": {
+                    "type": "string",
+                    "description": "The city that you want to get the weather for."
+                }, 
+                "days": {
+                    "type": "integer",
+                    "description": "Number of days for the forecast."
+                }
+            },
+            "required": ["location", "days"]
+        }
+    }
+]
+
+messages = [
+    {'role': 'system', 'content': system_prompt.format(functions=tools)},
+    {'role': 'user', 'content': query}
+]
+
+inputs = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt").to(model.device)
+
+outputs = model.generate(inputs, max_new_tokens=512, do_sample=False, num_return_sequences=1, eos_token_id=tokenizer.eos_token_id)
+print(tokenizer.decode(outputs[0][len(inputs[0]):], skip_special_tokens=True))

model.properties

+# 模型编码
+modelCode=1558
+# 模型名称
+modelName=ToolACE_pytorch
+# 模型描述
+modelDescription=华为研发的工具调用模型
+# 应用场景
+appScenario=训练,推理,对话问答,电商,教育,广媒
+# 框架类型
+frameType=pytorch

scripts/data_process.py

+import json
+import re
+
+def is_function_call(text):
+    """
+    判断文本是否符合函数调用列表格式：
+    [Func1(arg="val"), Func2(...)]
+    """
+    text = text.strip()
+    if not (text.startswith('[') and text.endswith(']')):
+        return False
+
+    inner = text[1:-1].strip()
+    pattern = re.compile(r'([^\(\)]+)\(([^()]*)\)')
+    matches = pattern.findall(inner)
+    if not matches:
+        return False
+
+    replaced = pattern.sub('', inner)
+    replaced = replaced.replace(',', '').replace(' ', '')
+    return replaced == ''
+
+def parse_function_call_list(text):
+    """
+    解析函数调用字符串为JSON数组格式
+    示例：
+    [SEC Filings(identifier="AAPL"), United States Away from Home Mobility API(string="2025-05-17")]
+    转换成：
+    [
+      {"name": "SEC Filings", "arguments": {"identifier": "AAPL"}},
+      {"name": "United States Away from Home Mobility API", "arguments": {"string": "2025-05-17"}}
+    ]
+    """
+    inner = text.strip()[1:-1].strip()
+    pattern = re.compile(r'([^\(\)]+)\((.*?)\)')
+    matches = pattern.findall(inner)
+
+    functions = []
+    for func_name, args_str in matches:
+        func_name = func_name.strip()
+
+        args = {}
+        if args_str.strip():
+            # 支持多个参数，形如 key="value", key2="value2"
+            parts = re.split(r',\s*(?=\w+=)', args_str)
+            for part in parts:
+                key_val = part.split('=', 1)
+                if len(key_val) == 2:
+                    key = key_val[0].strip()
+                    val = key_val[1].strip()
+                    if (val.startswith('"') and val.endswith('"')) or (val.startswith("'") and val.endswith("'")):
+                        val = val[1:-1]
+                    args[key] = val
+
+        functions.append({
+            "name": func_name,
+            "arguments": args
+        })
+
+    return functions
+
+def convert_conversation(conversations):
+    converted = []
+
+    for message in conversations:
+        role = message.get("from", "")
+        value = message.get("value", "")
+
+        if role == "user":
+            converted.append({
+                "from": "human",
+                "value": value
+            })
+
+        elif role == "assistant":
+            # 判断是否函数调用
+            if is_function_call(value):
+                parsed_funcs = parse_function_call_list(value)
+                converted.append({
+                    "from": "function_call",
+                    "value": json.dumps(parsed_funcs, ensure_ascii=False)
+                })
+            else:
+                converted.append({
+                    "from": "gpt",
+                    "value": value
+                })
+
+        elif role == "tool":
+            converted.append({
+                "from": "observation",
+                "value": value
+            })
+
+    return converted
+
+def transform_data(data):
+    """
+    对整体data进行转换，保留system，转换conversations
+    """
+    result = []
+    for item in data:
+        system_text = item.get("system", "")
+        conversations = item.get("conversations", [])
+
+        converted_conversations = convert_conversation(conversations)
+
+        result.append({
+            "system": system_text,
+            "conversations": converted_conversations
+        })
+
+    return result
+
+def main(input_file, output_file):
+    with open(input_file, "r", encoding="utf-8") as f:
+        data = json.load(f)
+
+    transformed = transform_data(data)
+
+    with open(output_file, "w", encoding="utf-8") as f:
+        json.dump(transformed, f, ensure_ascii=False, indent=2)
+
+
+if __name__ == "__main__":
+    from argparse import ArgumentParser
+    
+    parser = ArgumentParser()
+    
+    parser.add_argument("--input_file", type=str)
+
+    parser.add_argument("--output_file", type=str)
+    
+    args = parser.parse_args()
+
+    main(args.input_file, args.output_file)
+
+