permissions issue fix

Dockerfile

@@ -2,13 +2,15 @@
 FROM python:3.10.9
 
 # Copy the current directory contents into the container at .
-COPY . .
+COPY ./app ./app
 
 # Set the working directory to /
-WORKDIR /
+WORKDIR /trans
+
+EXPOSE 7860
 
 # Install requirements.txt 
 RUN pip install --no-cache-dir --upgrade -r /requirements.txt
 
 # Start the FastAPI app on port 7860, the default port expected by Spaces
-CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]
+CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "7860"]

app.py

@@ -1,35 +1,17 @@
-from pydantic import BaseModel
-
-from .ConfigEnv import config
+from fastapi import FastAPI
+from typing import List
+import torch
+from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
+from IndicTransToolkit import IndicProcessor
 from fastapi.middleware.cors import CORSMiddleware
 
-from langchain.llms import Clarifai
-from langchain.chains import LLMChain
-from langchain.prompts import PromptTemplate
-from TextGen import app
-
-
-class Generate(BaseModel):
-    text: str
-
-
-def generate_text(prompt: str):
-    if prompt == "":
-        return {"detail": "Please provide a prompt."}
-    else:
-        prompt = PromptTemplate(template=prompt, input_variables=["Prompt"])
-        llm = Clarifai(
-            pat=config.CLARIFAI_PAT,
-            user_id=config.USER_ID,
-            app_id=config.APP_ID,
-            model_id=config.MODEL_ID,
-            model_version_id=config.MODEL_VERSION_ID,
-        )
-        llmchain = LLMChain(prompt=prompt, llm=llm)
-        llm_response = llmchain.run({"Prompt": prompt})
-        return Generate(text=llm_response)
+import os
 
+os.environ["HF_HOME"] = "/.cache"
+# Initialize FastAPI
+app = FastAPI()
 
+# Add CORS middleware
 app.add_middleware(
     CORSMiddleware,
     allow_origins=["*"],
@@ -38,17 +20,67 @@ app.add_middleware(
     allow_headers=["*"],
 )
 
+# Initialize models and processors
+model = AutoModelForSeq2SeqLM.from_pretrained(
+    "ai4bharat/indictrans2-en-indic-1B", trust_remote_code=True
+)
+tokenizer = AutoTokenizer.from_pretrained(
+    "ai4bharat/indictrans2-en-indic-1B", trust_remote_code=True
+)
+ip = IndicProcessor(inference=True)
+DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
+model = model.to(DEVICE)
+
 
-@app.get("/", tags=["Home"])
-def api_home():
-    return {"detail": "Welcome to FastAPI TextGen Tutorial!"}
+def translate_text(sentences: List[str], target_lang: str):
+    try:
+        src_lang = "eng_Latn"
+        batch = ip.preprocess_batch(sentences, src_lang=src_lang, tgt_lang=target_lang)
+        inputs = tokenizer(
+            batch,
+            truncation=True,
+            padding="longest",
+            return_tensors="pt",
+            return_attention_mask=True,
+        ).to(DEVICE)
 
+        with torch.no_grad():
+            generated_tokens = model.generate(
+                **inputs,
+                use_cache=True,
+                min_length=0,
+                max_length=256,
+                num_beams=5,
+                num_return_sequences=1,
+            )
 
-@app.post(
-    "/api/generate",
-    summary="Generate text from prompt",
-    tags=["Generate"],
-    response_model=Generate,
-)
-def inference(input_prompt: str):
-    return generate_text(prompt=input_prompt)
+        with tokenizer.as_target_tokenizer():
+            generated_tokens = tokenizer.batch_decode(
+                generated_tokens.detach().cpu().tolist(),
+                skip_special_tokens=True,
+                clean_up_tokenization_spaces=True,
+            )
+
+        translations = ip.postprocess_batch(generated_tokens, lang=target_lang)
+        return {
+            "translations": translations,
+            "source_language": src_lang,
+            "target_language": target_lang,
+        }
+    except Exception as e:
+        raise Exception(f"Translation failed: {str(e)}")
+
+
+# FastAPI routes
+@app.get("/health")
+async def health_check():
+    return {"status": "healthy"}
+
+
+@app.post("/translate")
+async def translate_endpoint(sentences: List[str], target_lang: str):
+    try:
+        result = translate_text(sentences=sentences, target_lang=target_lang)
+        return result
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))

app/main.py

@@ -0,0 +1,79 @@
+from fastapi import FastAPI, HTTPException
+from typing import List
+from pydantic import BaseModel
+from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
+from IndicTransToolkit import IndicProcessor
+from fastapi.middleware.cors import CORSMiddleware
+
+app = FastAPI()
+
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],
+    allow_credentials=True,
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+model = AutoModelForSeq2SeqLM.from_pretrained(
+    "ai4bharat/indictrans2-en-indic-1B", trust_remote_code=True
+)
+tokenizer = AutoTokenizer.from_pretrained(
+    "ai4bharat/indictrans2-en-indic-1B", trust_remote_code=True
+)
+
+ip = IndicProcessor(inference=True)
+DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
+model = model.to(DEVICE)
+
+
+def translate_text(sentences: List[str], target_lang: str):
+    try:
+        src_lang = "eng_Latn"
+        batch = ip.preprocess_batch(sentences, src_lang=src_lang, tgt_lang=target_lang)
+        inputs = tokenizer(
+            batch,
+            truncation=True,
+            padding="longest",
+            return_tensors="pt",
+            return_attention_mask=True,
+        ).to(DEVICE)
+
+        with torch.no_grad():
+            generated_tokens = model.generate(
+                **inputs,
+                use_cache=True,
+                min_length=0,
+                max_length=256,
+                num_beams=5,
+                num_return_sequences=1,
+            )
+
+        with tokenizer.as_target_tokenizer():
+            generated_tokens = tokenizer.batch_decode(
+                generated_tokens.detach().cpu().tolist(),
+                skip_special_tokens=True,
+            )
+
+        return generated_tokens
+    except Exception as e:
+        return str(e)
+
+
+@app.get("/")
+def read_root():
+    return {"Hello": "World"}
+
+
+class TranslateRequest(BaseModel):
+    sentences: List[str]
+    target_lang: str
+
+
+@app.post("/translate/")
+def translate(request: TranslateRequest):
+    try:
+        result = translate_text(request.sentences, request.target_lang)
+        return result
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=str(e))

requirements.txt

@@ -1,7 +1,7 @@
-fastapi==0.99.1
+fastapi
 uvicorn
-requests
-pydantic==1.10.12
-langchain
-clarifai
-Pillow
+torch
+transformers
+git+https://github.com/VarunGumma/IndicTransToolkit.git
+python-multipart
+pydantic