Create app.py

app.py

@@ -0,0 +1,50 @@
+import os
+from fastapi import FastAPI
+from pydantic import BaseModel
+from transformers import pipeline, AutoTokenizer, AutoModelForCausalLM
+from langchain.llms import HuggingFacePipeline
+from langchain import PromptTemplate, LLMChain
+
+# — Model setup (small enough to CPU-serve in a Space) —
+MODEL_ID = "bigcode/starcoder2-1b"
+tokenizer = AutoTokenizer.from_pretrained(MODEL_ID)
+model = AutoModelForCausalLM.from_pretrained(MODEL_ID)
+
+# wrap in a HF pipeline and LangChain LLM
+pipe = pipeline(
+    "text-generation",
+    model=model,
+    tokenizer=tokenizer,
+    max_new_tokens=64,
+    temperature=0.2,
+    top_p=0.95,
+    do_sample=False,
+)
+llm = HuggingFacePipeline(pipeline=pipe)
+
+# define a simple prompt → chain
+prompt = PromptTemplate(
+    input_variables=["description"],
+    template=(
+        "### Convert English description to an Emmet abbreviation\n"
+        "Description: {description}\n"
+        "Emmet:"
+    ),
+)
+chain = LLMChain(llm=llm, prompt=prompt)
+
+# FastAPI app
+app = FastAPI()
+
+class Req(BaseModel):
+    description: str
+
+class Res(BaseModel):
+    emmet: str
+
+@app.post("/generate-emmet", response_model=Res)
+async def generate_emmet(req: Req):
+    raw = chain.run(req.description)
+    # take just the first line after the prompt
+    emmet = raw.strip().splitlines()[0]
+    return {"emmet": emmet}