Merge remote-tracking branch 'origin/dev_ty_06_end' into llm_framework

pilot/configs/config.py

@@ -18,6 +18,7 @@ class Config(metaclass=Singleton):
         """Initialize the Config class"""
 
         self.NEW_SERVER_MODE = False
+        self.SERVER_LIGHT_MODE = False
 
         # Gradio language version: en, zh
         self.LANGUAGE = os.getenv("LANGUAGE", "en")

pilot/connections/rdbms/py_study/study_data.py

@@ -1,10 +1,17 @@
+import json
 from pilot.common.sql_database import Database
 from pilot.configs.config import Config
 
 CFG = Config()
 
 if __name__ == "__main__":
-    connect = CFG.local_db.get_session("gpt-user")
+    # connect = CFG.local_db.get_session("gpt-user")
-    datas = CFG.local_db.run(connect, "SELECT * FROM users; ")
+    # datas = CFG.local_db.run(connect, "SELECT * FROM users; ")
+
+    # print(datas)
+
+    str = """{ "thoughts": "thought text", "sql": "SELECT COUNT(DISTINCT user_id) FROM transactions_order WHERE user_id IN (SELECT DISTINCT user_id FROM users WHERE country='China') AND create_time BETWEEN 20230101 AND 20230131" ,}"""
+
+    print(str.find("["))
+
 
-    print(datas)

pilot/model/llm_out/vicuna_base_llm.py

@@ -11,10 +11,10 @@ def generate_stream(
     """Fork from fastchat: https://github.com/lm-sys/FastChat/blob/main/fastchat/serve/inference.py"""
     prompt = params["prompt"]
     l_prompt = len(prompt)
+    prompt=  prompt.replace("ai:", "assistant:").replace("human:", "user:")
     temperature = float(params.get("temperature", 1.0))
     max_new_tokens = int(params.get("max_new_tokens", 2048))
     stop_str = params.get("stop", None)
-
     input_ids = tokenizer(prompt).input_ids
     output_ids = list(input_ids)
 

pilot/out_parser/base.py

@@ -113,25 +113,36 @@ class BaseOutputParser(ABC):
             ai_response = ai_response.replace("\n", " ")
             ai_response = ai_response.replace("\_", "_")
             ai_response = ai_response.replace("\*", "*")
+            ai_response = ai_response.replace("\t", "")
             print("un_stream ai response:", ai_response)
             return ai_response
         else:
             raise ValueError("Model server error!code=" + resp_obj_ex["error_code"])
 
+    def __illegal_json_ends(self, s):
+        temp_json = s
+        illegal_json_ends_1 = [", }", ",}"]
+        illegal_json_ends_2 = ", ]", ",]"
+        for illegal_json_end in illegal_json_ends_1:
+            temp_json = temp_json.replace(illegal_json_end, " }")
+        for illegal_json_end in illegal_json_ends_2:
+            temp_json = temp_json.replace(illegal_json_end, " ]")
+        return temp_json
+
     def __extract_json(self, s):
 
         temp_json = self.__json_interception(s, True)
         if not temp_json:
             temp_json = self.__json_interception(s)
         try:
-            json.loads(temp_json)
+            temp_json = self.__illegal_json_ends(temp_json)
             return temp_json
         except Exception as e:
             raise ValueError("Failed to find a valid json response！" + temp_json)
 
     def __json_interception(self, s, is_json_array: bool = False):
         if is_json_array:
-            i = s.index("[")
+            i = s.find("[")
             if i <0:
                 return None
             count = 1
@@ -145,7 +156,7 @@ class BaseOutputParser(ABC):
             assert count == 0
             return s[i: j + 1]
         else:
-            i = s.index("{")
+            i = s.find("{")
             if i <0:
                 return None
             count = 1
@@ -189,6 +200,7 @@ class BaseOutputParser(ABC):
                 .replace("\\n", " ")
                 .replace("\\", " ")
         )
+        cleaned_output = self.__illegal_json_ends(cleaned_output)
         return cleaned_output
 
     def parse_view_response(self, ai_text, data) -> str:

pilot/prompts/prompt_new.py

@@ -51,6 +51,9 @@ class PromptTemplate(BaseModel, ABC):
 
     need_historical_messages: bool = False
 
+    temperature: float = 0.6
+    max_new_tokens: int = 1024
+
     class Config:
         """Configuration for this pydantic object."""
 

pilot/scene/base_chat.py

@@ -48,8 +48,6 @@ CFG = Config()
 class BaseChat(ABC):
     chat_scene: str = None
     llm_model: Any = None
-    temperature: float = 0.6
-    max_new_tokens: int = 1024
     # By default, keep the last two rounds of conversation records as the context
     chat_retention_rounds: int = 1
 
@@ -118,8 +116,8 @@ class BaseChat(ABC):
         payload = {
             "model": self.llm_model,
             "prompt": self.generate_llm_text(),
-            "temperature": float(self.temperature),
+            "temperature": float(self.prompt_template.temperature),
-            "max_new_tokens": int(self.max_new_tokens),
+            "max_new_tokens": int(self.prompt_template.max_new_tokens),
             "stop": self.prompt_template.sep,
         }
         return payload
@@ -128,6 +126,7 @@ class BaseChat(ABC):
         # TODO Retry when server connection error
         payload = self.__call_base()
 
+
         self.skip_echo_len = len(payload.get("prompt").replace("</s>", " ")) + 11
         logger.info(f"Requert: \n{payload}")
         ai_response_text = ""

pilot/scene/chat_dashboard/template/sales_report/dashboard.json

@@ -3,7 +3,7 @@
   "name": "sale_report",
   "introduce": "",
   "layout": "TODO",
-  "supported_chart_type":["HeatMap","sheet", "LineChart", "PieChart", "BarChart", "Scatterplot", "IndicatorValue", "Table"],
+  "supported_chart_type":["FacetChart", "GaugeChart", "RadarChart", "Sheet", "LineChart", "PieChart", "BarChart", "PointChart", "KeyMetrics"],
   "key_metrics":[],
   "trends": []
 }

pilot/scene/chat_db/auto_execute/example.py

@@ -11,11 +11,11 @@ EXAMPLES = [
                 "data": {
                     "content": """{
 							\"thoughts\": \"thought text\",
-							\"sql\": \"SELECT city FROM users where user_name='test1'\",
+							\"sql\": \"SELECT city FROM user where user_name='test1'\",
 						}""",
                     "example": True,
-                },
+                }
-            },
+            }
         ]
     },
     {
@@ -26,13 +26,13 @@ EXAMPLES = [
                 "data": {
                     "content": """{
 							\"thoughts\": \"thought text\",
-							\"sql\": \"SELECT b.* FROM users a  LEFT JOIN tran_order b ON a.user_name=b.user_name  where a.city='成都'\",
+							\"sql\": \"SELECT b.* FROM user a  LEFT JOIN tran_order b ON a.user_name=b.user_name  where a.city='成都'\",
 						}""",
                     "example": True,
-                },
+                }
-            },
+            }
         ]
-    },
+    }
 ]
 
 sql_data_example = ExampleSelector(

pilot/scene/chat_db/auto_execute/out_parser.py

@@ -35,15 +35,16 @@ class DbChatOutputParser(BaseOutputParser):
         if len(data) <= 1:
             data.insert(0, ["result"])
         df = pd.DataFrame(data[1:], columns=data[0])
-        if not CFG.NEW_SERVER_MODE:
+        if not CFG.NEW_SERVER_MODE and not CFG.SERVER_LIGHT_MODE:
             table_style = """<style> 
                 table{border-collapse:collapse;width:100%;height:80%;margin:0 auto;float:center;border: 1px solid #007bff; background-color:#333; color:#fff}th,td{border:1px solid #ddd;padding:3px;text-align:center}th{background-color:#C9C3C7;color: #fff;font-weight: bold;}tr:nth-child(even){background-color:#444}tr:hover{background-color:#444}
              </style>"""
             html_table = df.to_html(index=False, escape=False)
             html = f"<html><head>{table_style}</head><body>{html_table}</body></html>"
         else:
-            html = df.to_html(index=False, escape=False, sparsify=False)
+            html_table = df.to_html(index=False, escape=False, sparsify=False)
-            html = "".join(html.split())
+            table_str = "".join(html_table.split())
+            html = f"""<div class="w-full overflow-auto">{table_str}</table></div>"""
 
         view_text = f"##### {str(speak)}" + "\n" + html.replace("\n", " ")
         return view_text

pilot/scene/chat_db/auto_execute/prompt.py

@@ -10,9 +10,8 @@ CFG = Config()
 
 PROMPT_SCENE_DEFINE = None
 
-
 _DEFAULT_TEMPLATE = """
-You are a SQL expert. Given an input question, create a syntactically correct {dialect} query.
+You are a SQL expert. Given an input question, create a syntactically correct {dialect} sql.
 
 Unless the user specifies in his question a specific number of examples he wishes to obtain, always limit your query to at most {top_k} results. 
 Use as few tables as possible when querying.
@@ -36,6 +35,11 @@ PROMPT_SEP = SeparatorStyle.SINGLE.value
 
 PROMPT_NEED_NEED_STREAM_OUT = False
 
+# Temperature is a configuration hyperparameter that controls the randomness of language model output.
+# A high temperature produces more unpredictable and creative results, while a low temperature produces more common and conservative output.
+# For example, if you adjust the temperature to 0.5, the model will usually generate text that is more predictable and less creative than if you set the temperature to 1.0.
+PROMPT_TEMPERATURE = 0.5
+
 prompt = PromptTemplate(
     template_scene=ChatScene.ChatWithDbExecute.value(),
     input_variables=["input", "table_info", "dialect", "top_k", "response"],
@@ -47,5 +51,6 @@ prompt = PromptTemplate(
         sep=PROMPT_SEP, is_stream_out=PROMPT_NEED_NEED_STREAM_OUT
     ),
     example_selector=sql_data_example,
+    temperature=PROMPT_TEMPERATURE
 )
 CFG.prompt_templates.update({prompt.template_scene: prompt})

pilot/scene/chat_execution/example.py

@@ -14,8 +14,8 @@ EXAMPLES = [
 							\"command\": {\"name\": \"command name\", \"args\": {\"arg name\": \"value\"}},
 						}""",
                     "example": True,
-                },
+                }
-            },
+            }
         ]
     },
     {
@@ -30,10 +30,10 @@ EXAMPLES = [
 							\"command\": {\"name\": \"command name\", \"args\": {\"arg name\": \"value\"}},
 						}""",
                     "example": True,
-                },
+                }
-            },
+            }
         ]
-    },
+    }
 ]
 
 plugin_example = ExampleSelector(examples_record=EXAMPLES, use_example=True)

pilot/server/dbgpt_server.py

@@ -103,7 +103,8 @@ if __name__ == "__main__":
         from pilot.server.llmserver import worker
         worker.start_check()
         CFG.NEW_SERVER_MODE = True
-
+    else:
+        CFG.SERVER_LIGHT_MODE = True
 
     import uvicorn
     uvicorn.run(app, host="0.0.0.0", port=args.port)