DeepSeek V3 0324：就在刚刚，DeepSeek悄悄更新，700行代码一气呵成！ 原创

!pip install llm
!llm install llm-mlx
!llm mlx download-model mlx-community/DeepSeek-V3-0324-4bit
1.
2.
3.

!llm chat -m mlx-community/DeepSeek-V3-0324-4bit 'Generate an SVG of a pelican riding a bicycle'
1.

!pip3 install openai
1.

from openai import OpenAI
import time

# 初始化计时
start_time = time.time()

# 初始化客户端
client = OpenAI(
    api_key="Your_api_key", 
    base_url="https://api.deepseek.com"
)

# 发送流式聊天请求
response = client.chat.completions.create(
    model="deepseek-chat",
    messages=[
        {"role": "system", "content": "You are a helpful assistant"},
        {"role": "user", "content": "How many r's are there in Strawberry"},
    ],
    stream=True
)

# 处理流式响应并收集指标
prompt_tokens = 0
generated_tokens = 0
full_response = ""

for chunk in response:
    if hasattr(chunk, "usage") and hasattr(chunk.usage, "prompt_tokens"):
        prompt_tokens = chunk.usage.prompt_tokens

    if hasattr(chunk, "choices") and hasattr(chunk.choices[0], "delta") and hasattr(chunk.choices[0].delta, "content"):
        content = chunk.choices[0].delta.content
        if content:
            generated_tokens += 1
            full_response += content
            print(content, end="", flush=True)

# 性能跟踪
end_time = time.time()
total_time = end_time - start_time

# 计算每秒token数
prompt_tps = prompt_tokens / total_time if prompt_tokens > 0else0
generation_tps = generated_tokens / total_time if generated_tokens > 0else0

# 输出指标
print("\n\n--- 性能指标 ---")
print(f"提示：{prompt_tokens} tokens, {prompt_tps:.3f} tokens-per-sec")
print(f"生成：{generated_tokens} tokens, {generation_tps:.3f} tokens-per-sec")
print(f"总时间：{total_time:.2f} 秒")
print(f"完整响应长度：{len(full_response)} 字符")
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.

!pip3 install openai

from openai import OpenAI
import time

# 记录开始时间
start_time = time.time()

client = OpenAI(api_key="Your_API_KEY", base_url="https://api.deepseek.com")

response = client.chat.completions.create(
    model="deepseek-chat",
    messages=[
        {"role": "system", "content": "You are a Website Developer"},
        {"role": "user", "content": "Code a modern small digital marketing Landing page"},
    ],
    stream=True
)

# 初始化变量以跟踪token和内容
prompt_tokens = 0
generated_tokens = 0
full_response = ""

# 处理流式响应
for chunk in response:
    if hasattr(chunk, "usage") and hasattr(chunk.usage, "prompt_tokens"):
        prompt_tokens = chunk.usage.prompt_tokens

    if hasattr(chunk, "choices") and hasattr(chunk.choices[0], "delta") and hasattr(chunk.choices[0].delta, "content"):
        content = chunk.choices[0].delta.content
        if content:
            generated_tokens += 1
            full_response += content
            print(content, end="", flush=True)

# 计算时间指标
end_time = time.time()
total_time = end_time - start_time

# 计算每秒token数
prompt_tps = prompt_tokens / total_time if prompt_tokens > 0else0
generation_tps = generated_tokens / total_time if generated_tokens > 0else0

# 输出指标
print("\n\n--- 性能指标 ---")
print(f"提示：{prompt_tokens} tokens, {prompt_tps:.3f} tokens-per-sec")
print(f"生成：{generated_tokens} tokens, {generation_tps:.3f} tokens-per-sec")
print(f"总时间：{total_time:.2f} 秒")
print(f"完整响应长度：{len(full_response)} 字符")
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
46.
47.
48.
49.
50.