Spaces:
Paused
Paused
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,15 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
from fastapi import FastAPI
|
| 2 |
from vllm import LLM, SamplingParams
|
| 3 |
from pydantic import BaseModel
|
| 4 |
from fastapi.responses import JSONResponse
|
| 5 |
import uvicorn
|
| 6 |
-
import os
|
| 7 |
-
os.environ["HF_HOME"] = "/data"
|
| 8 |
-
os.environ["TRANSFORMERS_CACHE"] = "/data"
|
| 9 |
-
os.environ["HF_HUB_CACHE"] = "/data"
|
| 10 |
|
| 11 |
# 1. 加载模型(会自动从 HuggingFace Hub 拉取)
|
| 12 |
-
llm = LLM(model="Qwen/Qwen2.5-Coder-7B-Instruct")
|
| 13 |
sampling = SamplingParams(temperature=0.2, max_tokens=1024)
|
| 14 |
|
| 15 |
app = FastAPI()
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
os.environ["HF_HOME"] = "/data"
|
| 3 |
+
os.environ["TRANSFORMERS_CACHE"] = "/data"
|
| 4 |
+
os.environ["HF_HUB_CACHE"] = "/data"
|
| 5 |
+
|
| 6 |
from fastapi import FastAPI
|
| 7 |
from vllm import LLM, SamplingParams
|
| 8 |
from pydantic import BaseModel
|
| 9 |
from fastapi.responses import JSONResponse
|
| 10 |
import uvicorn
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
|
| 12 |
# 1. 加载模型(会自动从 HuggingFace Hub 拉取)
|
| 13 |
+
llm = LLM(model="Qwen/Qwen2.5-Coder-7B-Instruct",download_dir="/data")
|
| 14 |
sampling = SamplingParams(temperature=0.2, max_tokens=1024)
|
| 15 |
|
| 16 |
app = FastAPI()
|