Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,5 +1,7 @@
|
|
| 1 |
import os
|
| 2 |
import time
|
|
|
|
|
|
|
| 3 |
import base64
|
| 4 |
from collections import defaultdict
|
| 5 |
from typing import List, Dict
|
|
@@ -999,7 +1001,8 @@ with gr.Blocks(
|
|
| 999 |
segs["rag_retrieve_done"] = (time.perf_counter() - t_r0) * 1000.0
|
| 1000 |
|
| 1001 |
# LLM (chat_with_clare must return 3 values)
|
| 1002 |
-
|
|
|
|
| 1003 |
answer, new_history, llm_stats = chat_with_clare(
|
| 1004 |
message=message,
|
| 1005 |
history=chat_history,
|
|
@@ -1012,8 +1015,15 @@ with gr.Blocks(
|
|
| 1012 |
cognitive_state=cognitive_state,
|
| 1013 |
rag_context=rag_context_text,
|
| 1014 |
)
|
| 1015 |
-
|
| 1016 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1017 |
|
| 1018 |
# merge llm_stats into perf
|
| 1019 |
perf = {
|
|
@@ -1239,17 +1249,18 @@ with gr.Blocks(
|
|
| 1239 |
{
|
| 1240 |
"experiment_id": experiment_id,
|
| 1241 |
"student_id": student_id,
|
| 1242 |
-
"event_type": "
|
| 1243 |
-
"timestamp":
|
| 1244 |
-
"latency_ms":
|
| 1245 |
-
"question":
|
| 1246 |
"answer": answer,
|
| 1247 |
"model_name": model_name_val,
|
| 1248 |
"language": resolved_lang,
|
| 1249 |
"learning_mode": mode_val,
|
| 1250 |
-
"
|
| 1251 |
}
|
| 1252 |
)
|
|
|
|
| 1253 |
except Exception as e:
|
| 1254 |
print("log_event error:", e)
|
| 1255 |
|
|
|
|
| 1 |
import os
|
| 2 |
import time
|
| 3 |
+
import json
|
| 4 |
+
|
| 5 |
import base64
|
| 6 |
from collections import defaultdict
|
| 7 |
from typing import List, Dict
|
|
|
|
| 1001 |
segs["rag_retrieve_done"] = (time.perf_counter() - t_r0) * 1000.0
|
| 1002 |
|
| 1003 |
# LLM (chat_with_clare must return 3 values)
|
| 1004 |
+
start_ts = time.time()
|
| 1005 |
+
|
| 1006 |
answer, new_history, llm_stats = chat_with_clare(
|
| 1007 |
message=message,
|
| 1008 |
history=chat_history,
|
|
|
|
| 1015 |
cognitive_state=cognitive_state,
|
| 1016 |
rag_context=rag_context_text,
|
| 1017 |
)
|
| 1018 |
+
|
| 1019 |
+
end_ts = time.time()
|
| 1020 |
+
latency_ms = (end_ts - start_ts) * 1000.0
|
| 1021 |
+
|
| 1022 |
+
# === optional: print profiled metrics locally ===
|
| 1023 |
+
try:
|
| 1024 |
+
print("[LLM_PROFILE] " + json.dumps(llm_stats, ensure_ascii=False))
|
| 1025 |
+
except Exception:
|
| 1026 |
+
pass
|
| 1027 |
|
| 1028 |
# merge llm_stats into perf
|
| 1029 |
perf = {
|
|
|
|
| 1249 |
{
|
| 1250 |
"experiment_id": experiment_id,
|
| 1251 |
"student_id": student_id,
|
| 1252 |
+
"event_type": "chat_turn",
|
| 1253 |
+
"timestamp": end_ts,
|
| 1254 |
+
"latency_ms": latency_ms,
|
| 1255 |
+
"question": message,
|
| 1256 |
"answer": answer,
|
| 1257 |
"model_name": model_name_val,
|
| 1258 |
"language": resolved_lang,
|
| 1259 |
"learning_mode": mode_val,
|
| 1260 |
+
"llm_stats": llm_stats,
|
| 1261 |
}
|
| 1262 |
)
|
| 1263 |
+
|
| 1264 |
except Exception as e:
|
| 1265 |
print("log_event error:", e)
|
| 1266 |
|