{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "import hopsworks\n",
    "from sentence_transformers import SentenceTransformer, InputExample, losses\n",
    "from torch.utils.data import DataLoader\n",
    "from sklearn.model_selection import train_test_split\n",
    "from dotenv import load_dotenv\n",
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2025-01-08 19:52:22,417 INFO: Closing external client and cleaning up certificates.\n",
      "Connection closed.\n",
      "2025-01-08 19:52:22,421 INFO: Initializing external client\n",
      "2025-01-08 19:52:22,421 INFO: Base URL: https://c.app.hopsworks.ai:443\n",
      "2025-01-08 19:52:23,548 INFO: Python Engine initialized.\n",
      "\n",
      "Logged in to project, explore it here https://c.app.hopsworks.ai:443/p/1158296\n"
     ]
    }
   ],
   "source": [
    "# Initialize Hopsworks connection\n",
    "load_dotenv()\n",
    "\n",
    "api_key = os.getenv(\"HOPSWORKS_API_KEY\")\n",
    "project = hopsworks.login(project=\"orestavf\", api_key_value=api_key)\n",
    "fs = project.get_feature_store()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Finished: Reading data from Hopsworks, using Hopsworks Feature Query Service (0.84s) \n"
     ]
    }
   ],
   "source": [
    "# Load preprocessed data\n",
    "feedback_fg = fs.get_feature_group(name=\"job_feedback\", version=1)\n",
    "feedback_df = feedback_fg.read()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Split into train and validation sets\n",
    "train_df, val_df = train_test_split(feedback_df, test_size=0.2, random_state=42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Prepare data for SentenceTransformer\n",
    "def prepare_examples(df):\n",
    "    examples = []\n",
    "    for _, row in df.iterrows():\n",
    "        examples.append(\n",
    "            InputExample(\n",
    "                texts=[row[\"resume_text\"], row[\"job_description\"]],\n",
    "                label=float(row[\"is_relevant\"])  # Convert to float for loss calculation\n",
    "            )\n",
    "        )\n",
    "    return examples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_examples = prepare_examples(train_df)\n",
    "val_examples = prepare_examples(val_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2025-01-08 19:25:05,476 INFO: Use pytorch device_name: cpu\n",
      "2025-01-08 19:25:05,477 INFO: Load pretrained SentenceTransformer: sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2\n"
     ]
    }
   ],
   "source": [
    "# Load pretrained SentenceTransformer\n",
    "model = SentenceTransformer('sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define DataLoader\n",
    "train_dataloader = DataLoader(train_examples, shuffle=True, batch_size=16)\n",
    "val_dataloader = DataLoader(val_examples, shuffle=False, batch_size=16)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define loss\n",
    "train_loss = losses.CosineSimilarityLoss(model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Configure training\n",
    "num_epochs = 3\n",
    "warmup_steps = int(len(train_dataloader) * num_epochs * 0.1)  # 10% of training as warmup"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "65a11878fdad456a94ae2e4d44e403a3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/3 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'train_runtime': 5.2094, 'train_samples_per_second': 2.879, 'train_steps_per_second': 0.576, 'train_loss': 0.27454523245493573, 'epoch': 3.0}\n",
      "2025-01-08 19:25:14,162 INFO: Save model to ./finetuned_model\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7bc7a5e2e56e4abe8bbf47e5ed251d6a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Computing widget examples:   0%|          | 0/1 [00:00<?, ?example/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a4218c62846f43c7be217513f8fd86de",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Computing widget examples:   0%|          | 0/1 [00:00<?, ?example/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Train the model\n",
    "model.fit(\n",
    "    train_objectives=[(train_dataloader, train_loss)],\n",
    "    evaluator=None,  # Add an evaluator if needed\n",
    "    epochs=num_epochs,\n",
    "    warmup_steps=warmup_steps,\n",
    "    output_path=\"./finetuned_model\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save the trained model locally\n",
    "#model.save(\"./finetuned_model\")\n",
    "#print(\"Model finetuned and saved locally!\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "from hsml.schema import Schema\n",
    "from hsml.model_schema import ModelSchema"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Define the Model Schema\n",
    "X_train_sample = train_df[[\"resume_text\", \"job_description\"]].sample(1).values  # Input example\n",
    "y_train_sample = train_df[\"is_relevant\"].sample(1).values  # Output example"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "input_schema = Schema(X_train_sample)\n",
    "output_schema = Schema(y_train_sample)\n",
    "model_schema = ModelSchema(input_schema=input_schema, output_schema=output_schema)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get Model Registry\n",
    "mr = project.get_model_registry()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Register the model in the Model Registry\n",
    "job_matching_model = mr.python.create_model(\n",
    "    name=\"job_matching_sentence_transformer\",\n",
    "    #metrics=metrics,\n",
    "    model_schema=model_schema,\n",
    "    input_example=X_train_sample,\n",
    "    description=\"Finetuned SentenceTransformer for job matching\",\n",
    "    version=1\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "78d7674e395848acb4586dd4dff1fee8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/6 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "ff6231d9469840229d992abb28a5740b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Uploading: 0.000%|          | 0/727 elapsed<00:00 remaining<?"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "53650ea045a24041a44626336d9a9c9b",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Uploading: 0.000%|          | 0/212 elapsed<00:00 remaining<?"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "21e688c69c7a46a4bb7485d1d088c887",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Uploading: 0.000%|          | 0/470637416 elapsed<00:00 remaining<?"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c68dc00bc2ae4804b73780d0f23e21ba",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Uploading: 0.000%|          | 0/242 elapsed<00:00 remaining<?"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "bb846d6b71b746749d59c0bf25779f61",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Uploading: 0.000%|          | 0/21034 elapsed<00:00 remaining<?"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "afda6046a57c42dca546be6defd13f11",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Uploading: 0.000%|          | 0/56 elapsed<00:00 remaining<?"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "df4c630a40954b9784a2a631be4ed4e3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Uploading: 0.000%|          | 0/1015 elapsed<00:00 remaining<?"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "4c6430695b7e43c0815fe278328e0448",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Uploading: 0.000%|          | 0/17082987 elapsed<00:00 remaining<?"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a1aa72dd0b344a26b22d8839aa12c4fb",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Uploading: 0.000%|          | 0/1512 elapsed<00:00 remaining<?"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2b599ffc95d4499b80673d873b6dfe7a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Uploading: 0.000%|          | 0/14763260 elapsed<00:00 remaining<?"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "fa11d5018fcb4ab094dac6c27128bcf5",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Uploading: 0.000%|          | 0/305 elapsed<00:00 remaining<?"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "7e37354a957d45b88b7a25007b3c7889",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Uploading: 0.000%|          | 0/6678 elapsed<00:00 remaining<?"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "f4e0c3e6d64b457bb01e5a0ac2162433",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Uploading: 0.000%|          | 0/216 elapsed<00:00 remaining<?"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model created, explore it at https://c.app.hopsworks.ai:443/p/1158296/models/job_matching_sentence_transformer/1\n",
      "Model registered in Hopsworks Model Registry!\n"
     ]
    }
   ],
   "source": [
    "# Save model artifacts to the Model Registry\n",
    "job_matching_model.save(\"./finetuned_model\")\n",
    "print(\"Model registered in Hopsworks Model Registry!\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2025-01-08 19:44:05,458 INFO: Save model to C:\\Users\\Filip\\AppData\\Local\\Temp\\tmpa217ndkp\n"
     ]
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "05f6bc89a66a4202b9f3b4b4fadee783",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Computing widget examples:   0%|          | 0/1 [00:00<?, ?example/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c1688de9db8d46b6ba410bcdcff839fd",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Computing widget examples:   0%|          | 0/1 [00:00<?, ?example/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c496a24b0d4044ffbbd9278a9f56996d",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "unigram.json:   0%|          | 0.00/14.8M [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "d4f0e6b478aa4a8dad8c798761289a8f",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "model.safetensors:   0%|          | 0.00/471M [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "e2a88564cf8b438bad836c593d4d78f6",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "Upload 3 LFS files:   0%|          | 0/3 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b302ff2e4cf74982bccdcf9d3a221240",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "tokenizer.json:   0%|          | 0.00/17.1M [00:00<?, ?B/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "'https://huggingface.co/forestav/job_matching_sentence_transformer/commit/7168a70785fae3fee6f5576b40a7556072ba31a2'"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Push the model to huggingface\n",
    "model.push_to_hub(\"forestav/job_matching_sentence_transformer\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "venv",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.12.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}