Spaces:
Sleeping
Sleeping
File size: 2,531 Bytes
933f161 169c067 933f161 169c067 933f161 169c067 933f161 169c067 933f161 |
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 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 |
from fastapi import FastAPI, HTTPException
from pydantic import BaseModel
from transformers import AutoTokenizer, AutoModelForCausalLM
from huggingface_hub import login
from transformers import StoppingCriteria, StoppingCriteriaList
import os
import torch
import uvicorn
class StopOnStrings(StoppingCriteria):
def __init__(self, tokenizer, stop_strings):
self.tokenizer = tokenizer
self.stop_ids = [tokenizer.encode(s, add_special_tokens=False) for s in stop_strings]
def __call__(self, input_ids, scores, **kwargs):
for stop_id in self.stop_ids:
if input_ids[0][-len(stop_id):].tolist() == stop_id:
return True
return False
login(os.getenv("HF_TOKEN"))
app = FastAPI(
title="VexaAI Model-Platform: Microsoft Phi-1.5",
description="Self-hosted AI-Model Microsoft Phi-1.5, powered by VexaAI.",
version="0.9"
)
model_name = "microsoft/phi-1_5"
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
model_name,
device_map="auto",
trust_remote_code=True,
torch_dtype=torch.float32
)
model.eval()
class GenerateRequest(BaseModel):
prompt: str
max_new_tokens: int = 512
temperature: float = 0.7
@app.post("/generate")
async def generate_text(request: GenerateRequest):
try:
inputs = tokenizer(request.prompt, return_tensors="pt").to(model.device)
with torch.no_grad():
stopping = StoppingCriteriaList([
StopOnStrings(tokenizer, ["\n\n", "###", "END"])
])
outputs = model.generate(
**inputs,
max_new_tokens=request.max_new_tokens,
temperature=request.temperature,
do_sample=True,
repetition_penalty=1.1,
pad_token_id=tokenizer.eos_token_id,
stopping_criteria=stopping
)
full_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
generated_text = full_text[len(tokenizer.decode(inputs["input_ids"][0], skip_special_tokens=True)):].strip()
return {"generated_text": generated_text}
except Exception as e:
raise HTTPException(status_code=500, detail=f"VexaAI Model-Platform: HTTP/S error: {str(e)}")
@app.get("/")
async def root():
return {"message": "To start generating text, use /generate."}
if __name__ == "__main__":
uvicorn.run(app, host="0.0.0.0", port=7860) |