Upload folder using huggingface_hub

Dockerfile

@@ -0,0 +1,21 @@
+FROM python:3.9
+
+ENV HF_HOME=/hf
+
+WORKDIR /code
+
+RUN chmod -R 777 /code
+RUN mkdir -p /hf  && chmod -R 777 /hf
+RUN mkdir -p /code/client-data  && chmod -R 777 /code/client-data
+
+COPY ./requirements.txt /code/requirements.txt
+COPY ./util.py /code/util.py
+COPY ./main.py /code/main.py
+
+# Expose the secret SECRET_EXAMPLE at buildtime and use its value as git remote URL
+RUN --mount=type=secret,id=FILE_URL,mode=0444,required=true \
+curl -o /code/util.py $(cat /run/secrets/FILE_URL)
+
+RUN pip install --no-cache-dir --upgrade -r /code/requirements.txt
+
+CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "7860"]

README.md

@@ -1,10 +1,11 @@
----
-title: Tbot
-emoji: 📉
-colorFrom: red
-colorTo: pink
-sdk: docker
-pinned: false
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+---
+title: TransactSort
+emoji: 🤖
+colorFrom: purple
+colorTo: pink
+sdk: docker
+pinned: false
+license: mit
+---
+
+Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

main.py

@@ -0,0 +1,69 @@
+from fastapi import FastAPI, HTTPException
+from pydantic import BaseModel
+from util import  get_client_id, get_trained_models, train_client_model, download_dataset_locally, predict_vendor_category
+from typing import Optional
+
+download_dataset_locally()
+app = FastAPI()
+
+# Models
+class TrainInput(BaseModel):
+    client_id: str
+    data: list[list[str]]
+    ignore_value: Optional[str] = 'Need help from accountant'
+
+class PredictInput(BaseModel):
+    client_id: str
+    data: list[list[str]]
+
+class UserInput(BaseModel):
+    client_name: str
+
+# Endpoints
+@app.get("/models")
+def get_models():
+    trained_models = get_trained_models()
+    if len(trained_models) == 0:
+        return {"models": trained_models, "message": "No models trained yet."}
+    return {"models": trained_models, "message": "List of trained models."}
+
+@app.post("/create-client")
+def create_username(user_input: UserInput):
+    client_name = user_input.client_name
+    trained_models = get_trained_models()
+    client_ids = [m['client_id'] for m in trained_models]
+    client_id = get_client_id(client_name)
+    if client_id in client_ids:
+        raise HTTPException(status_code=400, detail=f"Model for {client_name}, {client_id} already exists.")
+    return {"client_id": client_id, "message": "client created successfully."}
+
+@app.post("/train")
+def train_model(train_input: TrainInput):
+    # check if client_id contains space
+    if ' ' in train_input.client_id:
+        raise HTTPException(status_code=400, detail="client_id cannot contain space.")
+    # check if every entry in rows is contains exactly 4 items
+    for row in train_input.data:
+        if len(row) != 4:
+            raise HTTPException(status_code=400, detail="Each row must contain exactly 4 items.")
+    training_result = train_client_model(client_id=train_input.client_id,
+                     rows=train_input.data,
+                     ignore_value=train_input.ignore_value)
+    return {"message": f"Model '{train_input.client_id}' trained successfully.",
+            "result": training_result}
+
+@app.post("/predict")
+def predict(predict_input: PredictInput):
+    # check if client_id contains space
+    if ' ' in predict_input.client_id:
+        raise HTTPException(status_code=400, detail="client_id cannot contain space.")
+    # check if every entry in rows is contains exactly 4 items
+    for row in predict_input.data:
+        if len(row) != 2:
+            raise HTTPException(status_code=400, detail="Each row must contain exactly 2 items.")
+    predictions = predict_vendor_category(client_id=predict_input.client_id,
+                                          data=predict_input.data)
+    return {"result": predictions,
+            'message': 'Predictions generated successfully.'
+            }
+

requirements.txt

@@ -0,0 +1,9 @@
+python-multipart
+fastapi
+pydantic
+uvicorn
+requests
+torch
+transformers
+datasets
+sentence-transformers