Spaces:
Sleeping
Sleeping
Update server.py
Browse files
server.py
CHANGED
|
@@ -23,7 +23,7 @@ JSON_SOURCE = "database.json"
|
|
| 23 |
logging.basicConfig(level=logging.INFO)
|
| 24 |
logger = logging.getLogger(__name__)
|
| 25 |
|
| 26 |
-
app = FastAPI(title="Hebrew Smart Home Technician", version="11.
|
| 27 |
|
| 28 |
app.add_middleware(
|
| 29 |
CORSMiddleware,
|
|
@@ -86,10 +86,10 @@ def expand_query(query: str) -> List[str]:
|
|
| 86 |
return list(search_terms)
|
| 87 |
|
| 88 |
def simplify_schematic_text(text: str) -> str:
|
| 89 |
-
# ניקוי אגרסיבי והפיכה לעברית
|
| 90 |
text = text.replace("ESP32-DEVKITC-32D_ESP32-DEVKITC-30GPIO", "בקר ESP32")
|
| 91 |
text = text.replace("BOB_19002", "ממיר רמות")
|
| 92 |
text = text.replace("TLM-SH_board", "הלוח הראשי")
|
|
|
|
| 93 |
text = text.replace("NEXTION", "המסך")
|
| 94 |
text = text.replace("LCD1", "המסך")
|
| 95 |
text = text.replace("connects to", "מחובר אל")
|
|
@@ -132,13 +132,15 @@ def query_knowledge_base(query: str) -> str:
|
|
| 132 |
relevant_lines = []
|
| 133 |
target_terms = [t.lower() for t in search_terms if len(t) > 1]
|
| 134 |
for line in lines:
|
| 135 |
-
|
| 136 |
-
line_lower = line.lower()
|
| 137 |
-
if "connect" in line_lower and any(t in line_lower for t in target_terms):
|
| 138 |
relevant_lines.append(simplify_schematic_text(line.strip()))
|
| 139 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 140 |
if relevant_lines:
|
| 141 |
-
# הופך את הרשימה לפסקה קריאה יותר
|
| 142 |
return "הנה החיבורים שנמצאו:\n" + "\n".join(relevant_lines)
|
| 143 |
|
| 144 |
return raw_content
|
|
@@ -159,8 +161,8 @@ class Message(BaseModel):
|
|
| 159 |
|
| 160 |
class ChatCompletionRequest(BaseModel):
|
| 161 |
messages: List[Message]
|
| 162 |
-
max_tokens: Optional[int] =
|
| 163 |
-
temperature: Optional[float] = 0.
|
| 164 |
stream: Optional[bool] = False
|
| 165 |
|
| 166 |
@app.on_event("startup")
|
|
@@ -224,12 +226,12 @@ async def chat_completions(request: ChatCompletionRequest):
|
|
| 224 |
|
| 225 |
gen_kwargs = dict(
|
| 226 |
inputs=prompt_ids,
|
| 227 |
-
max_new_tokens=request.max_tokens,
|
| 228 |
-
temperature=0.4,
|
| 229 |
top_p=0.9,
|
| 230 |
do_sample=True,
|
| 231 |
pad_token_id=tokenizer.eos_token_id,
|
| 232 |
-
repetition_penalty=1.05
|
| 233 |
)
|
| 234 |
|
| 235 |
if request.stream:
|
|
|
|
| 23 |
logging.basicConfig(level=logging.INFO)
|
| 24 |
logger = logging.getLogger(__name__)
|
| 25 |
|
| 26 |
+
app = FastAPI(title="Hebrew Smart Home Technician", version="11.2.0")
|
| 27 |
|
| 28 |
app.add_middleware(
|
| 29 |
CORSMiddleware,
|
|
|
|
| 86 |
return list(search_terms)
|
| 87 |
|
| 88 |
def simplify_schematic_text(text: str) -> str:
|
|
|
|
| 89 |
text = text.replace("ESP32-DEVKITC-32D_ESP32-DEVKITC-30GPIO", "בקר ESP32")
|
| 90 |
text = text.replace("BOB_19002", "ממיר רמות")
|
| 91 |
text = text.replace("TLM-SH_board", "הלוח הראשי")
|
| 92 |
+
text = text.replace("StudentBoard", "הלוח הראשי")
|
| 93 |
text = text.replace("NEXTION", "המסך")
|
| 94 |
text = text.replace("LCD1", "המסך")
|
| 95 |
text = text.replace("connects to", "מחובר אל")
|
|
|
|
| 132 |
relevant_lines = []
|
| 133 |
target_terms = [t.lower() for t in search_terms if len(t) > 1]
|
| 134 |
for line in lines:
|
| 135 |
+
if "connect" in line.lower() and any(t in line.lower() for t in target_terms):
|
|
|
|
|
|
|
| 136 |
relevant_lines.append(simplify_schematic_text(line.strip()))
|
| 137 |
|
| 138 |
+
# Optimization: Limit to top 10 relevant lines to ensure completion
|
| 139 |
+
if len(relevant_lines) > 10:
|
| 140 |
+
relevant_lines = relevant_lines[:10]
|
| 141 |
+
relevant_lines.append("... (רשימה חלקית)")
|
| 142 |
+
|
| 143 |
if relevant_lines:
|
|
|
|
| 144 |
return "הנה החיבורים שנמצאו:\n" + "\n".join(relevant_lines)
|
| 145 |
|
| 146 |
return raw_content
|
|
|
|
| 161 |
|
| 162 |
class ChatCompletionRequest(BaseModel):
|
| 163 |
messages: List[Message]
|
| 164 |
+
max_tokens: Optional[int] = 1024 # Increased to prevent cut-off
|
| 165 |
+
temperature: Optional[float] = 0.3
|
| 166 |
stream: Optional[bool] = False
|
| 167 |
|
| 168 |
@app.on_event("startup")
|
|
|
|
| 226 |
|
| 227 |
gen_kwargs = dict(
|
| 228 |
inputs=prompt_ids,
|
| 229 |
+
max_new_tokens=request.max_tokens, # Now 1024
|
| 230 |
+
temperature=0.4,
|
| 231 |
top_p=0.9,
|
| 232 |
do_sample=True,
|
| 233 |
pad_token_id=tokenizer.eos_token_id,
|
| 234 |
+
repetition_penalty=1.05
|
| 235 |
)
|
| 236 |
|
| 237 |
if request.stream:
|