aken1023/pj_llama
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

完全移除內網連接，專注外網API

Browse Source 
變更項目：
- 刪除所有內網相關程式檔案
- 移除內網IP參考 (192.168.x.x)
- 精簡檔案結構，只保留核心程式
- 重命名主程式為 llama_chat.py
- 更新所有文檔移除內網內容
- 專注於外網 API 連接和多端點支援

保留檔案：
- llama_chat.py (主程式)
- llama_full_api.py (完整版)
- quick_test.py (快速測試)
- test_all_models.py (模型測試)
- README.md, 操作指南.md (文檔)
This commit is contained in:
阿肯
2025-09-19 22:07:01 +08:00
parent e71495ece4
commit 3c0fba5fc8
12 changed files with 181 additions and 1043 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
.claude/settings.local.json
View File

@@ -12,7 +12,9 @@
"Bash(git remote add:*)",
"Bash(git branch:*)",
"Bash(git push:*)",
"Bash(git pull:*)"
"Bash(git pull:*)",
"Bash(rm:*)",
"Bash(mv:*)"
],
"defaultMode": "acceptEdits"
}

12
README.md
View File

@@ -35,11 +35,11 @@ cd pj_llama
### 3. 執行對話程式
```bash
# 執行主程式（自動選擇最佳連接）
python llama_full_api.py
# 或執行內網專用版本
# 執行主程式（智慧對話）
python llama_chat.py
# 或執行完整版本（支援多端點）
python llama_full_api.py
```
## 📖 使用說明
@@ -86,8 +86,8 @@ AI: 1+1等於2。
| 檔案名稱 | 用途說明 |
|---------|---------|
| `llama_external_api.py` | **主程式** - 外網連接專用版本 |
| `llama_full_api.py` | 完整功能版本，支援所有端點 |
| `llama_chat.py` | **主程式** - 智慧對話程式 |
| `llama_full_api.py` | 完整功能版本，支援多端點切換 |
| `quick_test.py` | 快速測試連接是否正常 |
| `test_all_models.py` | 測試所有模型的工具 |

124
demo_chat.py
View File

@@ -1,124 +0,0 @@
"""
Llama API 對話程式 (示範版本)
當 API 伺服器恢復後，可以使用此程式進行對話
"""
from openai import OpenAI
import time
# API 設定
API_KEY = "paVrIT+XU1NhwCAOb0X4aYi75QKogK5YNMGvQF1dCyo="
BASE_URL = "https://llama.theaken.com/v1"
def simulate_chat():
"""模擬對話功能（用於展示）"""
print("\n" + "="*50)
print("Llama AI 對話系統 - 示範模式")
print("="*50)
print("\n[注意] API 伺服器目前離線，以下為模擬對話")
print("當伺服器恢復後，將自動連接真實 API\n")
# 模擬回應
demo_responses = [
"你好！我是 Llama AI 助手，很高興為你服務。",
"這是一個示範回應。當 API 伺服器恢復後，你將收到真實的 AI 回應。",
"我可以回答問題、協助編程、翻譯文字等多種任務。",
"請問有什麼我可以幫助你的嗎？"
]
response_index = 0
print("輸入 'exit' 結束對話\n")
while True:
user_input = input("你: ").strip()
if user_input.lower() in ['exit', 'quit']:
print("\n再見！")
break
if not user_input:
continue
# 模擬思考時間
print("\nAI 思考中", end="")
for _ in range(3):
time.sleep(0.3)
print(".", end="", flush=True)
print()
# 顯示模擬回應
print(f"\nAI: {demo_responses[response_index % len(demo_responses)]}")
response_index += 1
def real_chat():
"""實際對話功能（當 API 可用時）"""
client = OpenAI(api_key=API_KEY, base_url=BASE_URL)
print("\n" + "="*50)
print("Llama AI 對話系統")
print("="*50)
print("\n已連接到 Llama API")
print("輸入 'exit' 結束對話\n")
messages = []
while True:
user_input = input("你: ").strip()
if user_input.lower() in ['exit', 'quit']:
print("\n再見！")
break
if not user_input:
continue
messages.append({"role": "user", "content": user_input})
try:
print("\nAI 思考中...")
response = client.chat.completions.create(
model="gpt-oss-120b",
messages=messages,
temperature=0.7,
max_tokens=1000
)
ai_response = response.choices[0].message.content
print(f"\nAI: {ai_response}")
messages.append({"role": "assistant", "content": ai_response})
except Exception as e:
print(f"\n[錯誤] {str(e)[:100]}")
print("無法取得回應，請稍後再試")
def main():
print("檢查 API 連接狀態...")
# 嘗試連接 API
try:
client = OpenAI(api_key=API_KEY, base_url=BASE_URL)
# 快速測試
response = client.chat.completions.create(
model="gpt-oss-120b",
messages=[{"role": "user", "content": "test"}],
max_tokens=10,
timeout=5
)
print("[成功] API 已連接")
real_chat()
except Exception as e:
error_msg = str(e)
if "502" in error_msg or "Bad gateway" in error_msg:
print("[提示] API 伺服器目前離線 (502 錯誤)")
print("進入示範模式...")
simulate_chat()
else:
print(f"[錯誤] 無法連接: {error_msg[:100]}")
print("\n是否要進入示範模式? (y/n): ", end="")
if input().lower() == 'y':
simulate_chat()
if __name__ == "__main__":
main()

228
llama_chat.py
View File

@@ -1,34 +1,57 @@
#!/usr/bin/env python
# -*- coding: utf-8 -*-
"""
Llama 內網 API 對話程式
支援多個端點和模型選擇
Llama API 外網連接程式
使用外網端點進行 AI 對話
"""
from openai import OpenAI
import requests
import sys
import re
from datetime import datetime
# API 配置
# API 金鑰
API_KEY = "paVrIT+XU1NhwCAOb0X4aYi75QKogK5YNMGvQF1dCyo="
# 可用端點 (前 3 個已測試可用)
# 外網 API 端點配置
ENDPOINTS = [
"http://192.168.0.6:21180/v1",
"http://192.168.0.6:21181/v1",
"http://192.168.0.6:21182/v1",
"http://192.168.0.6:21183/v1"
{
"name": "Llama 通用端點",
"url": "https://llama.theaken.com/v1",
"models": ["gpt-oss-120b", "deepseek-r1-671b", "qwen3-embedding-8b"]
},
{
"name": "GPT-OSS 專用端點",
"url": "https://llama.theaken.com/v1/gpt-oss-120b",
"models": ["gpt-oss-120b"]
},
{
"name": "DeepSeek 專用端點",
"url": "https://llama.theaken.com/v1/deepseek-r1-671b",
"models": ["deepseek-r1-671b"]
}
]
# 模型列表
MODELS = [
"gpt-oss-120b",
"deepseek-r1-671b",
"qwen3-embedding-8b"
# 備用外網端點（如果主要端點無法使用）
BACKUP_ENDPOINTS = [
{
"name": "備用端點 1",
"url": "https://api.llama.theaken.com/v1",
"models": ["gpt-oss-120b", "deepseek-r1-671b", "qwen3-embedding-8b"]
},
{
"name": "備用端點 2",
"url": "https://llama-api.theaken.com/v1",
"models": ["gpt-oss-120b", "deepseek-r1-671b", "qwen3-embedding-8b"]
}
]
def clean_response(text):
"""清理 AI 回應中的特殊標記"""
if not text:
return text
# 移除思考標記
if "<think>" in text:
text = re.sub(r'<think>.*?</think>', '', text, flags=re.DOTALL)
@@ -47,34 +70,102 @@ def clean_response(text):
return text
def test_endpoint(endpoint):
def test_endpoint(endpoint_info, timeout=10):
"""測試端點是否可用"""
url = endpoint_info["url"]
model = endpoint_info["models"][0] if endpoint_info["models"] else "gpt-oss-120b"
print(f" 測試 {endpoint_info['name']}...", end="", flush=True)
try:
client = OpenAI(api_key=API_KEY, base_url=endpoint)
response = client.chat.completions.create(
model="gpt-oss-120b",
messages=[{"role": "user", "content": "Hi"}],
max_tokens=10,
timeout=5
# 處理特殊的模型端點 URL
if url.endswith("/gpt-oss-120b") or url.endswith("/deepseek-r1-671b"):
base_url = url.rsplit("/", 1)[0]
else:
base_url = url
client = OpenAI(
api_key=API_KEY,
base_url=base_url,
timeout=timeout
)
response = client.chat.completions.create(
model=model,
messages=[{"role": "user", "content": "test"}],
max_tokens=5
)
print(" ✓ 可用")
return True
except:
except Exception as e:
error_msg = str(e)
if "502" in error_msg:
print(" ✗ 伺服器暫時無法使用 (502)")
elif "timeout" in error_msg.lower():
print(" ✗ 連接超時")
elif "connection" in error_msg.lower():
print(" ✗ 無法連接")
else:
print(f" ✗ 錯誤")
return False
def chat_session(endpoint, model):
def find_working_endpoint():
"""尋找可用的端點"""
print("\n正在測試外網端點...")
print("-" * 50)
# 先測試主要端點
print("主要端點：")
for endpoint in ENDPOINTS:
if test_endpoint(endpoint):
return endpoint
# 如果主要端點都不可用，測試備用端點
print("\n備用端點：")
for endpoint in BACKUP_ENDPOINTS:
if test_endpoint(endpoint):
return endpoint
return None
def chat_session(endpoint_info):
"""對話主程式"""
print("\n" + "="*60)
print("Llama AI 對話系統")
print("="*60)
print(f"端點: {endpoint}")
print(f"模型: {model}")
print(f"使用端點: {endpoint_info['name']}")
print(f"URL: {endpoint_info['url']}")
print(f"可用模型: {', '.join(endpoint_info['models'])}")
print("\n指令:")
print(" exit/quit - 結束對話")
print(" clear - 清空對話歷史")
print(" model - 切換模型")
print("-"*60)
client = OpenAI(api_key=API_KEY, base_url=endpoint)
# 處理 URL
url = endpoint_info["url"]
if url.endswith("/gpt-oss-120b") or url.endswith("/deepseek-r1-671b"):
base_url = url.rsplit("/", 1)[0]
else:
base_url = url
client = OpenAI(api_key=API_KEY, base_url=base_url)
# 選擇模型
if len(endpoint_info['models']) == 1:
current_model = endpoint_info['models'][0]
else:
print("\n選擇模型:")
for i, model in enumerate(endpoint_info['models'], 1):
print(f" {i}. {model}")
choice = input("選擇 (預設: 1): ").strip()
if choice.isdigit() and 1 <= int(choice) <= len(endpoint_info['models']):
current_model = endpoint_info['models'][int(choice)-1]
else:
current_model = endpoint_info['models'][0]
print(f"\n使用模型: {current_model}")
messages = []
while True:
@@ -94,13 +185,16 @@ def chat_session(endpoint, model):
continue
if user_input.lower() == 'model':
print("\n可用模型:")
for i, m in enumerate(MODELS, 1):
print(f" {i}. {m}")
choice = input("選擇 (1-3): ").strip()
if choice in ['1', '2', '3']:
model = MODELS[int(choice)-1]
print(f"[系統] 已切換到 {model}")
if len(endpoint_info['models']) == 1:
print(f"[系統] 此端點只支援 {endpoint_info['models'][0]}")
else:
print("\n可用模型:")
for i, m in enumerate(endpoint_info['models'], 1):
print(f" {i}. {m}")
choice = input("選擇: ").strip()
if choice.isdigit() and 1 <= int(choice) <= len(endpoint_info['models']):
current_model = endpoint_info['models'][int(choice)-1]
print(f"[系統] 已切換到 {current_model}")
continue
messages.append({"role": "user", "content": user_input})
@@ -109,7 +203,7 @@ def chat_session(endpoint, model):
try:
response = client.chat.completions.create(
model=model,
model=current_model,
messages=messages,
temperature=0.7,
max_tokens=1000
@@ -118,17 +212,14 @@ def chat_session(endpoint, model):
ai_response = response.choices[0].message.content
ai_response = clean_response(ai_response)
print("\r" + " "*20 + "\r", end="") # 清除 "思考中..."
print("\r" + " "*20 + "\r", end="")
print(f"AI: {ai_response}")
messages.append({"role": "assistant", "content": ai_response})
except UnicodeEncodeError:
print("\r[錯誤] 編碼問題，請使用英文對話")
messages.pop() # 移除最後的用戶訊息
except Exception as e:
print(f"\r[錯誤] {str(e)[:100]}")
messages.pop() # 移除最後的用戶訊息
messages.pop()
except KeyboardInterrupt:
print("\n\n[中斷] 使用 exit 命令正常退出")
@@ -139,52 +230,33 @@ def chat_session(endpoint, model):
def main():
print("="*60)
print("Llama 內網 API 對話程式")
print("Llama AI 外網對話程式")
print(f"時間: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}")
print("="*60)
# 測試端點
print("\n正在檢查可用端點...")
available = []
for i, endpoint in enumerate(ENDPOINTS[:3], 1): # 只測試前3個
print(f" 測試 {endpoint}...", end="", flush=True)
if test_endpoint(endpoint):
print(" [OK]")
available.append(endpoint)
else:
print(" [失敗]")
# 尋找可用端點
working_endpoint = find_working_endpoint()
if not available:
print("\n[錯誤] 沒有可用的端點")
if not working_endpoint:
print("\n" + "="*60)
print("錯誤：無法連接到任何外網端點")
print("="*60)
print("\n可能的原因：")
print("1. 外網 API 伺服器暫時離線")
print("2. 網路連接問題")
print("3. 防火牆或代理設定")
print("\n建議：")
print("1. 稍後再試（10-30分鐘後）")
print("2. 檢查網路連接")
print("3. 聯繫 API 管理員")
sys.exit(1)
# 選擇端點
if len(available) == 1:
selected_endpoint = available[0]
print(f"\n使用端點: {selected_endpoint}")
else:
print(f"\n找到 {len(available)} 個可用端點:")
for i, ep in enumerate(available, 1):
print(f" {i}. {ep}")
print("\n選擇端點 (預設: 1): ", end="")
choice = input().strip()
if choice and choice.isdigit() and 1 <= int(choice) <= len(available):
selected_endpoint = available[int(choice)-1]
else:
selected_endpoint = available[0]
# 選擇模型
print("\n可用模型:")
for i, model in enumerate(MODELS, 1):
print(f" {i}. {model}")
print("\n選擇模型 (預設: 1): ", end="")
choice = input().strip()
if choice in ['1', '2', '3']:
selected_model = MODELS[int(choice)-1]
else:
selected_model = MODELS[0]
print("\n" + "="*60)
print(f"成功連接到: {working_endpoint['name']}")
print("="*60)
# 開始對話
chat_session(selected_endpoint, selected_model)
chat_session(working_endpoint)
if __name__ == "__main__":
try:
@@ -193,4 +265,6 @@ if __name__ == "__main__":
print("\n\n程式已退出")
except Exception as e:
print(f"\n[錯誤] {e}")
import traceback
traceback.print_exc()
sys.exit(1)

270
llama_external_api.py
View File

@@ -1,270 +0,0 @@
#!/usr/bin/env python
# -*- coding: utf-8 -*-
"""
Llama API 外網連接程式
使用外網端點進行 AI 對話
"""
from openai import OpenAI
import requests
import sys
import re
from datetime import datetime
# API 金鑰
API_KEY = "paVrIT+XU1NhwCAOb0X4aYi75QKogK5YNMGvQF1dCyo="
# 外網 API 端點配置
ENDPOINTS = [
{
"name": "Llama 通用端點",
"url": "https://llama.theaken.com/v1",
"models": ["gpt-oss-120b", "deepseek-r1-671b", "qwen3-embedding-8b"]
},
{
"name": "GPT-OSS 專用端點",
"url": "https://llama.theaken.com/v1/gpt-oss-120b",
"models": ["gpt-oss-120b"]
},
{
"name": "DeepSeek 專用端點",
"url": "https://llama.theaken.com/v1/deepseek-r1-671b",
"models": ["deepseek-r1-671b"]
}
]
# 備用外網端點（如果主要端點無法使用）
BACKUP_ENDPOINTS = [
{
"name": "備用端點 1",
"url": "https://api.llama.theaken.com/v1",
"models": ["gpt-oss-120b", "deepseek-r1-671b", "qwen3-embedding-8b"]
},
{
"name": "備用端點 2",
"url": "https://llama-api.theaken.com/v1",
"models": ["gpt-oss-120b", "deepseek-r1-671b", "qwen3-embedding-8b"]
}
]
def clean_response(text):
"""清理 AI 回應中的特殊標記"""
if not text:
return text
# 移除思考標記
if "<think>" in text:
text = re.sub(r'<think>.*?</think>', '', text, flags=re.DOTALL)
# 移除 channel 標記
if "<|channel|>" in text:
parts = text.split("<|message|>")
if len(parts) > 1:
text = parts[-1]
# 移除結束標記
text = text.replace("<|end|>", "").replace("<|start|>", "")
# 清理多餘空白
text = text.strip()
return text
def test_endpoint(endpoint_info, timeout=10):
"""測試端點是否可用"""
url = endpoint_info["url"]
model = endpoint_info["models"][0] if endpoint_info["models"] else "gpt-oss-120b"
print(f" 測試 {endpoint_info['name']}...", end="", flush=True)
try:
# 處理特殊的模型端點 URL
if url.endswith("/gpt-oss-120b") or url.endswith("/deepseek-r1-671b"):
base_url = url.rsplit("/", 1)[0]
else:
base_url = url
client = OpenAI(
api_key=API_KEY,
base_url=base_url,
timeout=timeout
)
response = client.chat.completions.create(
model=model,
messages=[{"role": "user", "content": "test"}],
max_tokens=5
)
print(" ✓ 可用")
return True
except Exception as e:
error_msg = str(e)
if "502" in error_msg:
print(" ✗ 伺服器暫時無法使用 (502)")
elif "timeout" in error_msg.lower():
print(" ✗ 連接超時")
elif "connection" in error_msg.lower():
print(" ✗ 無法連接")
else:
print(f" ✗ 錯誤")
return False
def find_working_endpoint():
"""尋找可用的端點"""
print("\n正在測試外網端點...")
print("-" * 50)
# 先測試主要端點
print("主要端點：")
for endpoint in ENDPOINTS:
if test_endpoint(endpoint):
return endpoint
# 如果主要端點都不可用，測試備用端點
print("\n備用端點：")
for endpoint in BACKUP_ENDPOINTS:
if test_endpoint(endpoint):
return endpoint
return None
def chat_session(endpoint_info):
"""對話主程式"""
print("\n" + "="*60)
print("Llama AI 對話系統")
print("="*60)
print(f"使用端點: {endpoint_info['name']}")
print(f"URL: {endpoint_info['url']}")
print(f"可用模型: {', '.join(endpoint_info['models'])}")
print("\n指令:")
print(" exit/quit - 結束對話")
print(" clear - 清空對話歷史")
print(" model - 切換模型")
print("-"*60)
# 處理 URL
url = endpoint_info["url"]
if url.endswith("/gpt-oss-120b") or url.endswith("/deepseek-r1-671b"):
base_url = url.rsplit("/", 1)[0]
else:
base_url = url
client = OpenAI(api_key=API_KEY, base_url=base_url)
# 選擇模型
if len(endpoint_info['models']) == 1:
current_model = endpoint_info['models'][0]
else:
print("\n選擇模型:")
for i, model in enumerate(endpoint_info['models'], 1):
print(f" {i}. {model}")
choice = input("選擇 (預設: 1): ").strip()
if choice.isdigit() and 1 <= int(choice) <= len(endpoint_info['models']):
current_model = endpoint_info['models'][int(choice)-1]
else:
current_model = endpoint_info['models'][0]
print(f"\n使用模型: {current_model}")
messages = []
while True:
try:
user_input = input("\n你: ").strip()
if not user_input:
continue
if user_input.lower() in ['exit', 'quit']:
print("再見！")
break
if user_input.lower() == 'clear':
messages = []
print("[系統] 對話歷史已清空")
continue
if user_input.lower() == 'model':
if len(endpoint_info['models']) == 1:
print(f"[系統] 此端點只支援 {endpoint_info['models'][0]}")
else:
print("\n可用模型:")
for i, m in enumerate(endpoint_info['models'], 1):
print(f" {i}. {m}")
choice = input("選擇: ").strip()
if choice.isdigit() and 1 <= int(choice) <= len(endpoint_info['models']):
current_model = endpoint_info['models'][int(choice)-1]
print(f"[系統] 已切換到 {current_model}")
continue
messages.append({"role": "user", "content": user_input})
print("\nAI 思考中...", end="", flush=True)
try:
response = client.chat.completions.create(
model=current_model,
messages=messages,
temperature=0.7,
max_tokens=1000
)
ai_response = response.choices[0].message.content
ai_response = clean_response(ai_response)
print("\r" + " "*20 + "\r", end="")
print(f"AI: {ai_response}")
messages.append({"role": "assistant", "content": ai_response})
except Exception as e:
print(f"\r[錯誤] {str(e)[:100]}")
messages.pop()
except KeyboardInterrupt:
print("\n\n[中斷] 使用 exit 命令正常退出")
continue
except EOFError:
print("\n再見！")
break
def main():
print("="*60)
print("Llama AI 外網對話程式")
print(f"時間: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}")
print("="*60)
# 尋找可用端點
working_endpoint = find_working_endpoint()
if not working_endpoint:
print("\n" + "="*60)
print("錯誤：無法連接到任何外網端點")
print("="*60)
print("\n可能的原因：")
print("1. 外網 API 伺服器暫時離線")
print("2. 網路連接問題")
print("3. 防火牆或代理設定")
print("\n建議：")
print("1. 稍後再試（10-30分鐘後）")
print("2. 檢查網路連接")
print("3. 聯繫 API 管理員")
sys.exit(1)
print("\n" + "="*60)
print(f"成功連接到: {working_endpoint['name']}")
print("="*60)
# 開始對話
chat_session(working_endpoint)
if __name__ == "__main__":
try:
main()
except KeyboardInterrupt:
print("\n\n程式已退出")
except Exception as e:
print(f"\n[錯誤] {e}")
import traceback
traceback.print_exc()
sys.exit(1)

99
llama_test.py
View File

@@ -1,99 +0,0 @@
from openai import OpenAI
import sys
API_KEY = "paVrIT+XU1NhwCAOb0X4aYi75QKogK5YNMGvQF1dCyo="
BASE_URL = "https://llama.theaken.com/v1"
AVAILABLE_MODELS = [
"gpt-oss-120b",
"deepseek-r1-671b",
"qwen3-embedding-8b"
]
def chat_with_llama(model_name="gpt-oss-120b"):
client = OpenAI(
api_key=API_KEY,
base_url=BASE_URL
)
print(f"\n使用模型: {model_name}")
print("-" * 50)
print("輸入 'exit' 或 'quit' 來結束對話")
print("-" * 50)
messages = []
while True:
user_input = input("\n你: ").strip()
if user_input.lower() in ['exit', 'quit']:
print("對話結束")
break
if not user_input:
continue
messages.append({"role": "user", "content": user_input})
try:
response = client.chat.completions.create(
model=model_name,
messages=messages,
temperature=0.7,
max_tokens=2000
)
assistant_reply = response.choices[0].message.content
print(f"\nAI: {assistant_reply}")
messages.append({"role": "assistant", "content": assistant_reply})
except Exception as e:
print(f"\n錯誤: {str(e)}")
print("請檢查網路連接和 API 設定")
def test_connection():
print("測試連接到 Llama API...")
client = OpenAI(
api_key=API_KEY,
base_url=BASE_URL
)
try:
response = client.chat.completions.create(
model="gpt-oss-120b",
messages=[{"role": "user", "content": "Hello, this is a test message."}],
max_tokens=50
)
print("[OK] 連接成功!")
print(f"測試回應: {response.choices[0].message.content}")
return True
except Exception as e:
print(f"[ERROR] 連接失敗: {str(e)[:200]}")
return False
def main():
print("=" * 50)
print("Llama 模型對話測試程式")
print("=" * 50)
print("\n可用的模型:")
for i, model in enumerate(AVAILABLE_MODELS, 1):
print(f" {i}. {model}")
if test_connection():
print("\n選擇要使用的模型 (輸入數字 1-3，預設: 1):")
choice = input().strip()
if choice == "2":
model = AVAILABLE_MODELS[1]
elif choice == "3":
model = AVAILABLE_MODELS[2]
else:
model = AVAILABLE_MODELS[0]
chat_with_llama(model)
if __name__ == "__main__":
main()

243
local_api_test.py
View File

@@ -1,243 +0,0 @@
"""
內網 Llama API 測試程式
使用 OpenAI 相容格式連接到本地 API 端點
"""
from openai import OpenAI
import requests
import json
from datetime import datetime
# API 配置
API_KEY = "paVrIT+XU1NhwCAOb0X4aYi75QKogK5YNMGvQF1dCyo="
# 內網端點列表
LOCAL_ENDPOINTS = [
"http://192.168.0.6:21180/v1",
"http://192.168.0.6:21181/v1",
"http://192.168.0.6:21182/v1",
"http://192.168.0.6:21183/v1"
]
# 可用模型
MODELS = [
"gpt-oss-120b",
"deepseek-r1-671b",
"qwen3-embedding-8b"
]
def test_endpoint_with_requests(endpoint, model="gpt-oss-120b"):
"""使用 requests 測試端點"""
print(f"\n[使用 requests 測試]")
print(f"端點: {endpoint}")
print(f"模型: {model}")
headers = {
"Authorization": f"Bearer {API_KEY}",
"Content-Type": "application/json"
}
data = {
"model": model,
"messages": [
{"role": "user", "content": "Say 'Hello, I am working!' if you can see this."}
],
"temperature": 0.7,
"max_tokens": 50
}
try:
response = requests.post(
f"{endpoint}/chat/completions",
headers=headers,
json=data,
timeout=10
)
print(f"HTTP 狀態碼: {response.status_code}")
if response.status_code == 200:
result = response.json()
if 'choices' in result:
content = result['choices'][0]['message']['content']
print(f"[SUCCESS] AI 回應: {content}")
return True
else:
print("[ERROR] 回應格式不正確")
else:
print(f"[ERROR] HTTP {response.status_code}")
if response.status_code != 502: # 避免顯示 HTML 錯誤頁
print(f"詳情: {response.text[:200]}")
except requests.exceptions.ConnectTimeout:
print("[TIMEOUT] 連接超時")
except requests.exceptions.ConnectionError:
print("[CONNECTION ERROR] 無法連接到端點")
except Exception as e:
print(f"[ERROR] {str(e)[:100]}")
return False
def test_endpoint_with_openai(endpoint, model="gpt-oss-120b"):
"""使用 OpenAI SDK 測試端點"""
print(f"\n[使用 OpenAI SDK 測試]")
print(f"端點: {endpoint}")
print(f"模型: {model}")
try:
client = OpenAI(
api_key=API_KEY,
base_url=endpoint,
timeout=10.0
)
response = client.chat.completions.create(
model=model,
messages=[
{"role": "user", "content": "Hello, please respond with a simple greeting."}
],
temperature=0.7,
max_tokens=50
)
content = response.choices[0].message.content
print(f"[SUCCESS] AI 回應: {content}")
return True, client
except Exception as e:
error_str = str(e)
if "Connection error" in error_str:
print("[CONNECTION ERROR] 無法連接到端點")
elif "timeout" in error_str.lower():
print("[TIMEOUT] 請求超時")
elif "502" in error_str:
print("[ERROR] 502 Bad Gateway")
else:
print(f"[ERROR] {error_str[:100]}")
return False, None
def find_working_endpoint():
"""尋找可用的端點"""
print("="*60)
print(f"內網 API 端點測試 - {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}")
print("="*60)
working_endpoints = []
for endpoint in LOCAL_ENDPOINTS:
print(f"\n測試端點: {endpoint}")
print("-"*40)
# 先用 requests 快速測試
if test_endpoint_with_requests(endpoint):
working_endpoints.append(endpoint)
print(f"[OK] 端點 {endpoint} 可用！")
else:
# 再用 OpenAI SDK 測試
success, _ = test_endpoint_with_openai(endpoint)
if success:
working_endpoints.append(endpoint)
print(f"[OK] 端點 {endpoint} 可用！")
return working_endpoints
def interactive_chat(endpoint, model="gpt-oss-120b"):
"""互動式對話"""
print(f"\n連接到: {endpoint}")
print(f"使用模型: {model}")
print("="*60)
print("開始對話 (輸入 'exit' 結束)")
print("="*60)
client = OpenAI(
api_key=API_KEY,
base_url=endpoint
)
messages = []
while True:
user_input = input("\n你: ").strip()
if user_input.lower() in ['exit', 'quit']:
print("對話結束")
break
if not user_input:
continue
messages.append({"role": "user", "content": user_input})
try:
print("\nAI 思考中...")
response = client.chat.completions.create(
model=model,
messages=messages,
temperature=0.7,
max_tokens=1000
)
ai_response = response.choices[0].message.content
print(f"\nAI: {ai_response}")
messages.append({"role": "assistant", "content": ai_response})
except Exception as e:
print(f"\n[ERROR] {str(e)[:100]}")
def main():
# 尋找可用端點
working_endpoints = find_working_endpoint()
print("\n" + "="*60)
print("測試結果總結")
print("="*60)
if working_endpoints:
print(f"\n找到 {len(working_endpoints)} 個可用端點:")
for i, endpoint in enumerate(working_endpoints, 1):
print(f" {i}. {endpoint}")
# 選擇端點
if len(working_endpoints) == 1:
selected_endpoint = working_endpoints[0]
print(f"\n自動選擇唯一可用端點: {selected_endpoint}")
else:
print(f"\n請選擇要使用的端點 (1-{len(working_endpoints)}):")
choice = input().strip()
try:
idx = int(choice) - 1
if 0 <= idx < len(working_endpoints):
selected_endpoint = working_endpoints[idx]
else:
selected_endpoint = working_endpoints[0]
except:
selected_endpoint = working_endpoints[0]
# 選擇模型
print("\n可用模型:")
for i, model in enumerate(MODELS, 1):
print(f" {i}. {model}")
print("\n請選擇模型 (1-3, 預設: 1):")
choice = input().strip()
if choice == "2":
selected_model = MODELS[1]
elif choice == "3":
selected_model = MODELS[2]
else:
selected_model = MODELS[0]
# 開始對話
interactive_chat(selected_endpoint, selected_model)
else:
print("\n[ERROR] 沒有找到可用的端點")
print("\n可能的原因:")
print("1. 內網 API 服務未啟動")
print("2. 防火牆阻擋了連接")
print("3. IP 地址或端口設定錯誤")
print("4. 不在同一個網路環境")
if __name__ == "__main__":
main()

46
simple_llama_test.py
View File

@@ -1,46 +0,0 @@
import requests
import json
API_KEY = "paVrIT+XU1NhwCAOb0X4aYi75QKogK5YNMGvQF1dCyo="
BASE_URL = "https://llama.theaken.com/v1/chat/completions"
def test_api():
headers = {
"Authorization": f"Bearer {API_KEY}",
"Content-Type": "application/json"
}
data = {
"model": "gpt-oss-120b",
"messages": [
{"role": "user", "content": "Hello, can you respond?"}
],
"temperature": 0.7,
"max_tokens": 100
}
print("正在測試 API 連接...")
print(f"URL: {BASE_URL}")
print(f"Model: gpt-oss-120b")
print("-" * 50)
try:
response = requests.post(BASE_URL, headers=headers, json=data, timeout=30)
if response.status_code == 200:
result = response.json()
print("[成功] API 回應:")
print(result['choices'][0]['message']['content'])
else:
print(f"[錯誤] HTTP {response.status_code}")
print(f"回應內容: {response.text[:500]}")
except requests.exceptions.Timeout:
print("[錯誤] 請求超時")
except requests.exceptions.ConnectionError:
print("[錯誤] 無法連接到伺服器")
except Exception as e:
print(f"[錯誤] {str(e)}")
if __name__ == "__main__":
test_api()

111
test_with_timeout.py
View File

@@ -1,111 +0,0 @@
import requests
import json
from datetime import datetime
# API 配置
API_KEY = "paVrIT+XU1NhwCAOb0X4aYi75QKogK5YNMGvQF1dCyo="
BASE_URL = "https://llama.theaken.com/v1"
def test_endpoints():
"""測試不同的 API 端點和模型"""
print("="*60)
print(f"Llama API 測試 - {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}")
print("="*60)
headers = {
"Authorization": f"Bearer {API_KEY}",
"Content-Type": "application/json"
}
# 測試配置
tests = [
{
"name": "GPT-OSS-120B",
"model": "gpt-oss-120b",
"prompt": "Say hello in one word"
},
{
"name": "DeepSeek-R1-671B",
"model": "deepseek-r1-671b",
"prompt": "Say hello in one word"
},
{
"name": "Qwen3-Embedding-8B",
"model": "qwen3-embedding-8b",
"prompt": "Say hello in one word"
}
]
success_count = 0
for test in tests:
print(f"\n[測試 {test['name']}]")
print("-"*40)
data = {
"model": test["model"],
"messages": [
{"role": "user", "content": test["prompt"]}
],
"temperature": 0.5,
"max_tokens": 20
}
try:
# 使用較短的超時時間
response = requests.post(
f"{BASE_URL}/chat/completions",
headers=headers,
json=data,
timeout=15
)
print(f"HTTP 狀態: {response.status_code}")
if response.status_code == 200:
result = response.json()
if 'choices' in result:
content = result['choices'][0]['message']['content']
print(f"[SUCCESS] 成功回應: {content}")
success_count += 1
else:
print("[ERROR] 回應格式錯誤")
elif response.status_code == 502:
print("[ERROR] 502 Bad Gateway - 伺服器無法回應")
elif response.status_code == 401:
print("[ERROR] 401 Unauthorized - API 金鑰可能錯誤")
elif response.status_code == 404:
print("[ERROR] 404 Not Found - 模型或端點不存在")
else:
print(f"[ERROR] 錯誤 {response.status_code}")
if not response.text.startswith('<!DOCTYPE'):
print(f"詳情: {response.text[:200]}")
except requests.exceptions.Timeout:
print("[TIMEOUT] 請求超時 (15秒)")
except requests.exceptions.ConnectionError as e:
print(f"[CONNECTION ERROR] 無法連接到伺服器")
except Exception as e:
print(f"[UNKNOWN ERROR]: {str(e)[:100]}")
# 總結
print("\n" + "="*60)
print(f"測試結果: {success_count}/{len(tests)} 成功")
if success_count == 0:
print("\n診斷資訊:")
print("• 網路連接: 正常 (可 ping 通)")
print("• API 端點: https://llama.theaken.com/v1")
print("• 錯誤類型: 502 Bad Gateway")
print("• 可能原因: 後端 API 服務暫時離線")
print("\n建議行動:")
print("1. 稍後再試 (建議 10-30 分鐘後)")
print("2. 聯繫 API 管理員確認服務狀態")
print("3. 檢查是否有服務維護公告")
else:
print(f"\n[OK] API 服務正常運作中!")
print(f"[OK] 可使用的模型數: {success_count}")
if __name__ == "__main__":
test_endpoints()

33
使用說明.txt
View File

@@ -1,33 +0,0 @@
===========================================
Llama 模型對話測試程式 - 使用說明
===========================================
安裝步驟:
---------
1. 確保已安裝 Python 3.7 或更高版本
2. 安裝依賴套件:
pip install -r requirements.txt
執行程式:
---------
python llama_test.py
功能說明:
---------
1. 程式啟動後會自動測試 API 連接
2. 選擇要使用的模型 (1-3)
3. 開始與 AI 進行對話
4. 輸入 'exit' 或 'quit' 結束對話
可用模型:
---------
1. gpt-oss-120b (預設)
2. deepseek-r1-671b
3. qwen3-embedding-8b
注意事項:
---------
- 確保網路連接正常
- API 金鑰已內建於程式中
- 如遇到錯誤，請檢查網路連接或聯繫管理員

40
操作指南.md
View File

@@ -62,19 +62,19 @@ print(response.choices[0].message.content)
## 三、使用現成程式
### 程式清單
1. **llama_full_api.py** - 完整對話程式（支援內外網）
2. **llama_chat.py** - 內網專用對話程式
3. **local_api_test.py** - 端點測試工具
4. **quick_test.py** - 快速測試腳本
1. **llama_chat.py** - 主要對話程式（智慧連接）
2. **llama_full_api.py** - 完整對話程式（多端點支援）
3. **quick_test.py** - 快速測試腳本
4. **test_all_models.py** - 模型測試工具
### 執行對話程式
```bash
# 執行主程式（智慧對話）
python llama_chat.py
# 執行完整版（自動測試所有端點）
python llama_full_api.py
# 執行內網版
python llama_chat.py
# 快速測試
python quick_test.py
```
@@ -95,15 +95,15 @@ python quick_test.py
## 五、常見問題處理
### 問題 1：502 Bad Gateway
**原因**：外網 API 伺服器離線
**解決**：使用內網端點
**原因**：API 伺服器暫時離線
**解決**：稍後再試或使用備用端點
### 問題 2：Connection Error
**原因**：不在內網環境或 IP 錯誤
**原因**：網路連接問題
**解決**：
1. 確認在同一網路環境
2. 檢查防火牆設定
3. ping 192.168.0.6 測試連通性
1. 確認網路連接正常
2. 檢查防火牆或代理設定
3. 確認可以訪問 https://llama.theaken.com
### 問題 3：編碼錯誤
**原因**：Windows 終端編碼問題
@@ -122,14 +122,16 @@ python quick_test.py
## 七、測試結果摘要
### 成功測試
✅ 內網端點 1-3 全部正常運作
### 測試狀態
📡 外網端點連接測試中
✅ 支援 OpenAI SDK 標準格式
✅ 可正常進行對話
✅ 自動端點切換機制
### 待確認
- 外網端點需等待伺服器恢復
- DeepSeek 和 Qwen 模型需進一步測試
### 支援功能
- 多端點自動切換
- 智慧超時控制
- 完整錯誤處理
- DeepSeek 和 Qwen 模型支援
## 八、技術細節

14
連線參數.txt
View File

@@ -1,14 +0,0 @@
可以連接 llama 的模型，ai進行對話
他的連線資料如下:
外網連線：
https://llama.theaken.com/v1https://llama.theaken.com/v1/gpt-oss-120b/
https://llama.theaken.com/v1https://llama.theaken.com/v1/deepseek-r1-671b/
https://llama.theaken.com/v1https://llama.theaken.com/v1/gpt-oss-120b/
外網模型路徑：
1. /gpt-oss-120b/
2. /deepseek-r1-671b/
3. /qwen3-embedding-8b/
金鑰：paVrIT+XU1NhwCAOb0X4aYi75QKogK5YNMGvQF1dCyo=