Ollama对话脚本

import sys

import ollama

OLLAMA_HOST = "http://127.0.0.1:11434"

MODEL_NAME = "qwen3.5:2b"

try:

client = ollama.Client(host=OLLAMA_HOST)

client.show(MODEL_NAME)

except Exception as e:

print(f"连接失败: {e}")

sys.exit(1)

1. 严格保持线性的上下文结构，绝不在中间乱插数据

messages = \[\]

print("已对齐 Page Assist 优化参数。开始聊天：")

print("-" * 50)

while True:

try:

user_input = input("\n用户: ").strip()

if user_input.lower() in 'exit', 'quit': break

if not user_input: continue

messages.append({'role': 'user', 'content': user_input})

print("AI: ", end='', flush=True)

2. 显式传入 options 优化参数，对齐前端 WebUI 的标准配置

response_stream = client.chat(

model=MODEL_NAME,

messages=messages,

stream=True,

options={

"num_ctx": 4096, # 限制上下文大小，大幅提升响应速度

"low_vram": False, # 确保不向 CPU 妥协（如果显存够）

"temperature": 0.7 # 保持正常的发散度

}

)

full_response = ""

for chunk in response_stream:

content = chunk.message.content

if content:

print(content, end='', flush=True)

full_response += content

print()

messages.append({'role': 'assistant', 'content': full_response})

3. 严格的滑动窗口：超过限制时，必须两两成对裁剪，不能打破 user, assistant 的交替交叠

如果打破了交替，Ollama 缓存会当场报废，速度直接暴跌

if len(messages) > 10: # 保持最近 5 轮对话

messages = messages-10:

except KeyboardInterrupt:

print("\n退出")

break

except Exception as e:

print(f"\n错误: {e}")

break