fix and clean finetunning data collection

edenia · Mar 25, 2024 · b5fe5fa · b5fe5fa
1 parent 42e770f
commit b5fe5fa
Showing 1 changed file with 8 additions and 14 deletions.
diff --git a/analysis/ethereum_smart_contracts/GPT_finetuning_data_collection.ipynb b/analysis/ethereum_smart_contracts/GPT_finetuning_data_collection.ipynb
@@ -9,7 +9,9 @@
     "# Malicious Smart Contract Detection Training Dataset Collection Notebook\n",
     "\n",
     "This notebook collects smart contract creation bytecode and decompiled opcodes for malicious contract classification. \n",
-    "Benign contracts are gathered from Zettablock and malicious contracts from [Forta Network's labelled datasets github repo](\"https://github.com/forta-network/labelled-datasets\"). "
+    "Benign contracts are gathered from Zettablock and malicious contracts from [Forta Network's labelled datasets github repo](\"https://github.com/forta-network/labelled-datasets\").\n",
+    "\n",
+    "# Code provided by the Forta project"
    ]
   },
   {
@@ -54,7 +56,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "ALCHEMY_URL = \"https://eth-mainnet.g.alchemy.com/v2/f8hEqd_nKEpamacB-zXeWPv7R9QmWKPx\"\n",
+    "ALCHEMY_URL = \"https://eth-mainnet.g.alchemy.com/v2/XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX\"\n",
     "w3_eth = Web3(Web3.HTTPProvider(ALCHEMY_URL))\n",
     "TRACES = {}\n",
     "\n",
@@ -129,7 +131,7 @@
    "outputs": [],
    "source": [
     "def get_malicious_contracts() -> pd.DataFrame:\n",
-    "    data_path = '/data/forta/ethereum/text/malicious_data.pkl'\n",
+    "    data_path = '/data/forta/ethereum/text/finetuning/malicious_data.pkl'\n",
     "    malicious = None\n",
     "\n",
     "    if os.path.exists(data_path):\n",
@@ -157,7 +159,7 @@
    "outputs": [],
    "source": [
     "CONTRACT_DATA = {}\n",
-    "ETHERSCAN_API_KEY=\"SX7737W8M9DDTYHSXSV8XG8G945PMC9D4U\"\n",
+    "ETHERSCAN_API_KEY=\"XXXXXXXXXXXXXXXX\"\n",
     "\n",
     "def get_contract_transactions(contracts):\n",
     "    \"\"\"Get contract transaction info from Etherscan.\"\"\"\n",
@@ -211,7 +213,7 @@
     "'''\n",
     "\n",
     "def get_benign_contracts() -> pd.DataFrame:\n",
-    "    data_path = '/data/forta/ethereum/text/benign_data.pkl'\n",
+    "    data_path = '/data/forta/ethereum/text/finetuning/malicious_data.pkl'\n",
     "    benign = None\n",
     "\n",
     "    if os.path.exists(data_path):\n",
@@ -297,7 +299,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "dataset.fillna('').to_parquet('/data/forta/ethereum/text/malicious_contract_training_dataset_final.parquet', index=None)"
+    "dataset.fillna('').to_parquet('/data/forta/ethereum/text/finetuning/malicious_contract_training_dataset_final.parquet', index=None)"
    ]
   },
   {
@@ -309,14 +311,6 @@
    "source": [
     "dataset['malicious'].value_counts().plot(kind='pie', figsize=(7, 7))"
    ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "c9e10b1b-61bb-47fc-b608-2bc9f226b7f8",
-   "metadata": {},
-   "outputs": [],
-   "source": []
   }
  ],
  "metadata": {