File size: 1,750 Bytes
bd7fedf | 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 | from fastapi import FastAPI
from pydantic import BaseModel
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
MODEL_ID = "meta-llama/Meta-Llama-3.1-8B-Instruct"
app = FastAPI(title="Artist Description Generator")
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
model = AutoModelForCausalLM.from_pretrained(
MODEL_ID,
device_map="auto",
torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
load_in_4bit=torch.cuda.is_available()
)
class ArtistInput(BaseModel):
artist_name: str
country: str | None = None
genres: list[str] = []
top_tracks: list[str] = []
metrics: dict = {}
def build_prompt(data: ArtistInput) -> str:
return f"""
You are writing a factual artist description for a music analytics platform.
Rules:
- Use ONLY the provided data
- Do NOT invent awards, numbers, or events
- If data is missing, omit it
- Keep it concise and neutral
- 3–5 sentences maximum
Artist name: {data.artist_name}
Country: {data.country}
Genres: {", ".join(data.genres)}
Top tracks: {", ".join(data.top_tracks)}
Metrics: {data.metrics}
Artist description:
""".strip()
@app.post("/generate")
def generate_description(data: ArtistInput):
prompt = build_prompt(data)
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
with torch.no_grad():
output = model.generate(
**inputs,
max_new_tokens=180,
temperature=0.3,
top_p=0.9,
do_sample=True
)
text = tokenizer.decode(output[0], skip_special_tokens=True)
description = text.split("Artist description:")[-1].strip()
return {
"artist": data.artist_name,
"description": description
}
|