File size: 1,544 Bytes
57913ca
 
 
9a4add7
 
 
57913ca
9a4add7
57913ca
 
9a4add7
57913ca
 
 
 
 
 
 
9a4add7
57913ca
9a4add7
 
 
 
 
 
57913ca
 
 
9a4add7
57913ca
 
 
9a4add7
57913ca
 
 
 
9a4add7
57913ca
 
 
 
 
 
 
 
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
from fastapi import FastAPI
from pydantic import BaseModel
from transformers import pipeline, AutoTokenizer, AutoModelForCausalLM
from diffusers import StableDiffusionPipeline
import torch

app = FastAPI(title="Custom Content Generator API")

TEXT_MODEL = "gpt2"
IMAGE_MODEL = "runwayml/stable-diffusion-v1-5"
device = "cuda" if torch.cuda.is_available() else "cpu"

# Text pipeline
tokenizer = AutoTokenizer.from_pretrained(TEXT_MODEL)
model = AutoModelForCausalLM.from_pretrained(TEXT_MODEL)
text_pipe = pipeline("text-generation", model=model, tokenizer=tokenizer, device_map="auto" if device=="cuda" else None)

# Image pipeline (optional)
try:
    sd_pipe = StableDiffusionPipeline.from_pretrained(IMAGE_MODEL, torch_dtype=torch.float16 if device=="cuda" else torch.float32)
    sd_pipe = sd_pipe.to(device)
    has_sd = True
except Exception as e:
    sd_pipe = None
    has_sd = False

class TextReq(BaseModel):
    prompt: str
    max_len: int = 150

class ImageReq(BaseModel):
    prompt: str
    steps: int = 25

@app.post("/generate/text")
def gen_text(req: TextReq):
    out = text_pipe(req.prompt, max_length=req.max_len, do_sample=True)
    return {"generated": out[0]["generated_text"]}

@app.post("/generate/image")
def gen_image(req: ImageReq):
    if not has_sd:
        return {"error": "Image model not loaded."}
    img = sd_pipe(req.prompt, num_inference_steps=req.steps).images[0]
    # save image temporarily and return path or return bytes (for demo keep simple)
    img.save("out.png")
    return {"path": "/out.png"}