2 branches of generating examples.

demo/sample_generator.ipynb

@@ -73,29 +73,61 @@
     "            \"\"\"Given the following JSON example for a task type:\n",
     "            {raw_example}\n",
     "\n",
-    "            Provide a concise description of the task type.\n",
+    "            Provide a concise description of the task type, including the format and style of the output.\n",
     "\n",
     "            Format your response as follows:\n",
     "            Task Description: [Your description here]\n",
     "            \"\"\"\n",
     "        )\n",
     "\n",
-    "        self.examples_prompt = ChatPromptTemplate.from_template(\n",
-    "            \"\"\"Given the following task type description and JSON example:\n",
+    "        self.briefs_prompt = ChatPromptTemplate.from_template(\n",
+    "            \"\"\"Given the following task type description:\n",
     "            Task Description: {description}\n",
-    "            \n",
+    "\n",
+    "            Generate descriptions for 3 new examples with detailed attributes\n",
+    "            based on this task type.\n",
+    "\n",
+    "            Format your response as a valid YAML object with a single key\n",
+    "            'brief_descriptions' containing a YAML array of 3 objects, each\n",
+    "            with a 'description' field.\n",
+    "            \"\"\"\n",
+    "        )\n",
+    "\n",
+    "        self.examples_from_briefs_prompt = ChatPromptTemplate.from_template(\n",
+    "            \"\"\"Given the following task type description, brief description for new examples, \n",
+    "            and JSON example:\n",
+    "\n",
+    "            Task Description: {description}\n",
+    "            Brief Description: {brief_description}\n",
+    "\n",
     "            Example:\n",
     "            {raw_example}\n",
     "\n",
-    "            Generate 3 more input/output examples for this task type in the same JSON format.\n",
+    "            Generate 3 more input/output examples for this task type, based on the brief\n",
+    "            description, in the same JSON format.\n",
     "\n",
-    "            Format your response as a valid JSON object with a single key 'examples' containing a JSON array of 3 objects, each with 'input' and 'output' fields.\n",
+    "            Format your response as a valid JSON object with a single key 'examples' \n",
+    "            containing a JSON array of 3 objects, each with 'input' and 'output' fields.\n",
+    "            \"\"\"\n",
+    "        )\n",
+    "\n",
+    "        self.examples_prompt = ChatPromptTemplate.from_template(\n",
+    "            \"\"\"Given the following task type description, and input/output example:\n",
+    "            Task Description: {description}\n",
+    "            Example: {example}\n",
+    "\n",
+    "            Generate 3 new input/output examples for this task type.\n",
+    "\n",
+    "            Format your response as a valid JSON object with a single key 'examples' \n",
+    "            containing a JSON array of 3 objects, each with 'input' and 'output' fields.\n",
     "            \"\"\"\n",
     "        )\n",
     "\n",
     "        self.description_chain = self.description_prompt | self.model | self.output_parser\n",
+    "        self.briefs_chain = self.briefs_prompt | self.model | self.output_parser\n",
     "        # bind json_object to the model\n",
     "        json_model = self.model.bind(response_format={\"type\": \"json_object\"})\n",
+    "        self.examples_from_briefs_chain = self.examples_from_briefs_prompt | json_model | self.output_parser\n",
     "        self.examples_chain = self.examples_prompt | json_model | self.output_parser\n",
     "\n",
     "    def generate_description(self, raw_example_str):\n",
@@ -104,9 +136,17 @@
     "        })\n",
     "        return result.split(\"Task Description: \")[1].strip()\n",
     "\n",
-    "    def generate_examples(self, description, raw_example_str):\n",
-    "        result = self.examples_chain.invoke({\n",
+    "    def generate_briefs(self, description):\n",
+    "        result = self.briefs_chain.invoke({\n",
+    "            \"description\": description\n",
+    "        })\n",
+    "        # return result.split(\"Brief Description: \")[1].strip()\n",
+    "        return result\n",
+    "\n",
+    "    def generate_examples_from_briefs(self, description, brief_description, raw_example_str):\n",
+    "        result = self.examples_from_briefs_chain.invoke({\n",
     "            \"description\": description,\n",
+    "            \"brief_description\": brief_description,\n",
     "            \"raw_example\": raw_example_str\n",
     "        })\n",
     "\n",
@@ -119,6 +159,22 @@
     "            return json.loads(result)\n",
     "        except json.JSONDecodeError as e:\n",
     "            raise ValueError(f\"The generated examples are not in valid JSON format. Error: {str(e)} Result: {result}\")\n",
+    "        \n",
+    "    def generate_examples(self, description, example):\n",
+    "        result = self.examples_chain.invoke({\n",
+    "            \"description\": description,\n",
+    "            \"example\": example\n",
+    "        })\n",
+    "        \n",
+    "        try:\n",
+    "            result = result.strip()\n",
+    "            if result.startswith('```') and result.endswith('```'):\n",
+    "                result = result.strip('```').strip()\n",
+    "                if result.startswith('json'):\n",
+    "                    result = result.strip('json').strip()\n",
+    "            return json.loads(result)\n",
+    "        except json.JSONDecodeError as e:\n",
+    "            raise ValueError(f\"The generated examples are not in valid JSON format. Error: {str(e)} Result: {result}\")\n",
     "\n",
     "    def process(self, input_str):\n",
     "        try:\n",
@@ -135,15 +191,18 @@
     "            if not isinstance(data, dict) or 'input' not in data or 'output' not in data:\n",
     "                raise ValueError(\"Invalid input format. Expected an object with 'input' and 'output' fields.\")\n",
     "            \n",
-    "            description = self.generate_description(json.dumps(data))\n",
-    "            new_examples = self.generate_examples(description, json.dumps(data))\n",
-    "            \n",
+    "            description = self.generate_description(json.dumps(data, ensure_ascii=False))\n",
+    "            brief_description = self.generate_briefs(description)\n",
+    "            new_examples_from_briefs = self.generate_examples_from_briefs(description, brief_description, json.dumps(data, ensure_ascii=False))\n",
+    "            new_examples = self.generate_examples(description, json.dumps(data, ensure_ascii=False))\n",
+    "    \n",
     "            output = {\n",
     "                \"task_description\": description,\n",
-    "                \"additional_examples\": new_examples\n",
+    "                \"brief_description\": brief_description,\n",
+    "                \"additional_examples\": list(new_examples_from_briefs.values()) + list(new_examples.values())\n",
     "            }\n",
     "            \n",
-    "            return json.dumps(output, indent=2)\n",
+    "            return json.dumps(output, indent=2, ensure_ascii=False)\n",
     "        \n",
     "        except Exception as e:\n",
     "            raise RuntimeError(f\"An error occurred during processing: {str(e)}\")\n",