update

requirements.txt

@@ -46,7 +46,7 @@ tokenizers
 
 # camelot-py
 # llama_index
-byzerllm[saas]>=0.1.135
+byzerllm[saas]>=0.1.136
 patch
 diff_match_patch
 GitPython

src/auto_coder.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: auto-coder
-Version: 0.1.182
+Version: 0.1.183
 Summary: AutoCoder: AutoCoder
 Author: allwefantasy
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
@@ -26,7 +26,7 @@ Requires-Dist: tabulate
 Requires-Dist: jupyter_client
 Requires-Dist: prompt-toolkit
 Requires-Dist: tokenizers
-Requires-Dist: byzerllm[saas]>=0.1.135
+Requires-Dist: byzerllm[saas]>=0.1.136
 Requires-Dist: patch
 Requires-Dist: diff_match_patch
 Requires-Dist: GitPython

src/auto_coder.egg-info/requires.txt

@@ -16,7 +16,7 @@ tabulate
 jupyter_client
 prompt-toolkit
 tokenizers
-byzerllm[saas]>=0.1.135
+byzerllm[saas]>=0.1.136
 patch
 diff_match_patch
 GitPython

src/autocoder/auto_coder_rag.py

@@ -231,6 +231,17 @@ def main(input_args: Optional[List[str]] = None):
         help="Disable reordering of document segments after retrieval",
     )
 
+    serve_parser.add_argument(
+        "--disable_inference_enhance",
+        action="store_true",
+        help="Disable enhanced inference mode",
+    )
+    serve_parser.add_argument(
+        "--inference_deep_thought",
+        action="store_true",
+        help="Enable deep thought in inference mode",
+    )
+
     # Tools command
     tools_parser = subparsers.add_parser("tools", help="Various tools")
     tools_subparsers = tools_parser.add_subparsers(dest="tool", help="Available tools")

src/autocoder/chat_auto_coder.py

@@ -718,10 +718,11 @@ def get_completions(self, document, complete_event):
 
                 parser.coding()
                 current_word = parser.current_word()
-
-                for command in parser.get_sub_commands():
-                    if command.startswith(current_word):
-                        yield Completion(command, start_position=-len(current_word))
+
+                if len(new_text.strip()) == 0 or new_text.strip()=="/":
+                    for command in parser.get_sub_commands():
+                        if command.startswith(current_word):
+                            yield Completion(command, start_position=-len(current_word))
 
                 all_tags = parser.tags
 
@@ -796,6 +797,8 @@ def get_completions(self, document, complete_event):
                                 start_position=-len(name),
                                 display=f"{symbol.symbol_name} ({display_name}/{symbol.symbol_type})",
                             )
+
+                tags = [tag for tag in parser.tags]
 
                 if current_word.startswith("<"):
                     name = current_word[1:]
@@ -807,9 +810,7 @@ def get_completions(self, document, complete_event):
                                 )
                         elif tag.startswith(name):
                             yield Completion(tag, start_position=-len(current_word))
-
-                tags = [tag for tag in parser.tags]
-
+
                 if tags and tags[-1].start_tag == "<img>" and tags[-1].end_tag == "":
                     raw_file_name = tags[0].content
                     file_name = raw_file_name.strip()

src/autocoder/common/__init__.py

@@ -312,5 +312,8 @@ class AutoCoderArgs(pydantic.BaseModel):
     segment_ratio: Optional[float] = 0.2
     buff_ratio: Optional[float] = 0.1
 
+    disable_inference_enhance: Optional[bool] = False
+    inference_deep_thought: Optional[bool] = False
+
     class Config:
         protected_namespaces = ()

src/autocoder/rag/long_context_rag.py

@@ -492,7 +492,10 @@ def response_generator():
 
             if LLMComputeEngine is not None:
                 llm_compute_engine = LLMComputeEngine(
-                    llm=self.llm, inference_enhance=True, debug=False
+                    llm=self.llm,
+                    inference_enhance=not self.args.disable_inference_enhance,
+                    inference_deep_thought=self.args.inference_deep_thought,
+                    debug=False,
                 )
                 new_conversations = llm_compute_engine.process_conversation(
                     conversations, query, [doc.source_code for doc in relevant_docs]

src/autocoder/version.py

@@ -1 +1 @@
-__version__ = "0.1.182"
+__version__ = "0.1.183"