| """Confirm thinking is disabled and params are correct."""
|
| import os, json, requests, time
|
| os.chdir(os.path.dirname(os.path.abspath(__file__)))
|
| from backend.chat import ChatHandler
|
| from backend.memory import DatabaseManager
|
| from backend.config import config
|
|
|
|
|
| print("=== Current config ===")
|
| print(f" Model : {config.LLM_MODEL}")
|
| print(f" Thinking : {config.LLM_ENABLE_THINKING}")
|
| print(f" Temperature : {config.LLM_TEMPERATURE}")
|
| print(f" Top-p : {config.LLM_TOP_P}")
|
| print(f" Top-k : {config.LLM_TOP_K}")
|
| print(f" Presence penalty: {config.LLM_PRESENCE_PENALTY}")
|
| print(f" Max tokens : {config.LLM_MAX_TOKENS}")
|
|
|
|
|
| print("\n=== Live call ===")
|
| HEADERS = {"Content-Type": "application/json",
|
| "Authorization": f"Bearer {config.ATLASCLOUD_API_KEY}"}
|
| handler = ChatHandler.__new__(ChatHandler)
|
| handler.api_url = "https://api.atlascloud.ai/v1/chat/completions"
|
| handler.headers = HEADERS
|
| handler.db = None
|
|
|
| msgs = [{"role": "user", "content": "In one sentence: what is your motto?"}]
|
| payload = handler._build_payload(msgs, stream=True)
|
| print(f" Payload keys: {list(payload.keys())}")
|
| print(f" enable_thinking in payload: {payload.get('enable_thinking')}")
|
|
|
| t0 = time.time()
|
| r = requests.post(handler.api_url, headers=HEADERS, json=payload,
|
| stream=True, timeout=(30, 120))
|
| print(f" HTTP {r.status_code}")
|
|
|
| thinking_chars, content = 0, ""
|
| first_token_at = None
|
| for line in r.iter_lines():
|
| if not line: continue
|
| txt = line.decode()
|
| if not txt.startswith("data: "): continue
|
| raw = txt[6:].strip()
|
| if raw == "[DONE]": break
|
| try:
|
| d = json.loads(raw)
|
| delta = d["choices"][0].get("delta", {})
|
| rc = delta.get("reasoning_content", "")
|
| c = delta.get("content", "")
|
| if rc: thinking_chars += len(rc)
|
| if c:
|
| if first_token_at is None: first_token_at = time.time() - t0
|
| content += c
|
| except: pass
|
|
|
| elapsed = time.time() - t0
|
| print(f" Time to first token : {first_token_at:.1f}s")
|
| print(f" Total time : {elapsed:.1f}s")
|
| print(f" Thinking chars : {thinking_chars} {'(NONE - thinking OFF)' if not thinking_chars else '(THINKING IS STILL ON!)'}")
|
| print(f" Response : {content}")
|
|
|
