Validation (#1027)

* add validation script * update * change token count function * reorganize cells * Add unit tests * Add a printout for CPT * update question * Add questions * Fix lints * update format * update * nb source * add validation script * update * change token count function * reorganize cells * Add unit tests * Add a printout for CPT * update question * Add questions * Fix lints * update format * update * nb source * Remove license insert for validation notebook * Add validation utils * Minor cleanups (#858) * nits * logger * add log * lint * update utils/__init__.py to include extra validation functions * update notebook * update * update * Read UC delta table (#773) * initial commit * use databricks-sql to read delta table and convert to json * update * update * update * add mocked unittest * Fix lints * update * update * restructure code * Add timer for optimizing * Add db-connect * add wrapper * update * add install dbconnect * update * update * patch dbconnect to allow multiple return formats * update * add arrow * use compression * clean up * Add cluster rt check * Fix lints * remove patch.py for CI * update * update * updat * update * fix tests * fix lint * update * update * Add more tests * update * update * update * change to download_json * update * fix lints * Add decompressed option for arrow * format json to jsonl * Add comments * Make cf_collect_type global option * fix comments * fix lints * fix comments * Fix lints * change to use workspaceclient * Add CPT support * Rewire method assignment logic * Fix bug in stripping https * Add tests for rewired method assignment logic * Fix lints * Fix lints * Removed logger set_level * Remove pyspark. It conflicts with databricks-connect * Update the comment * skip cluster version check when cluster_id is serverless * Add use_serverless flag * update tests with use_serverless flag * Fix lints --------- Co-authored-by: Xiaohan Zhang <[email protected]> * Add download remote function to util * update * remove fused layernorm (#859) * update * update * update * update * update * update * update * update * update * Remove hardcoded combined.jsonl with a flag (#861) * Remove hardcoded combined.jsonl with a flag * update * change output_json_path output_json_folder --------- Co-authored-by: Xiaohan Zhang <[email protected]> * bump (#828) * Add dask and dataframe_to_mds * update * update * update * update * Add notebook * update * update * remove script and tests, keep notebook * update * update * update * update * Always initialize dist (#864) * fix dev * lint * remove gpu * updated notebook * remove scripts keep notebook * update notebook. rephrase. * update * Add response tokens * update * update * Disable MDSWrite, return token counts * Change plot settings * update notebook * update * update notebook * update * update notebook * update pip install link * Change done file location * Create the dest folder * update notebook * update --------- Co-authored-by: Xiaohan Zhang <[email protected]> Co-authored-by: xiaohanzhan-db <xiaohanzhan-db> Co-authored-by: Mihir Patel <[email protected]>
mosaicml · Mar 14, 2024 · 9fd91cf · 9fd91cf
1 parent 5090e13
commit 9fd91cf
Showing 1 changed file with 163 additions and 21 deletions.
diff --git a/notebooks/validate_and_tokenize_data.ipynb b/notebooks/validate_and_tokenize_data.ipynb
@@ -4,7 +4,10 @@
    "cell_type": "markdown",
    "metadata": {
     "application/vnd.databricks.v1+cell": {
-     "cellMetadata": {},
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
      "inputWidgets": {},
      "nuid": "f275a21b-47d4-472c-972b-e2a84a597db2",
      "showTitle": false,
@@ -54,7 +57,10 @@
    "cell_type": "markdown",
    "metadata": {
     "application/vnd.databricks.v1+cell": {
-     "cellMetadata": {},
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
      "inputWidgets": {},
      "nuid": "3d08a21c-9f5a-4ad2-af85-e016335cc53d",
      "showTitle": false,
@@ -173,6 +179,7 @@
     "import re\n",
     "import json\n",
     "import tempfile\n",
+    "import random\n",
     "import numpy as np\n",
     "import pandas as pd \n",
     "from collections import defaultdict\n",
@@ -193,7 +200,10 @@
    "cell_type": "markdown",
    "metadata": {
     "application/vnd.databricks.v1+cell": {
-     "cellMetadata": {},
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
      "inputWidgets": {},
      "nuid": "3a513cdd-967d-4a87-b56f-340053fa79cd",
      "showTitle": false,
@@ -208,7 +218,10 @@
    "cell_type": "markdown",
    "metadata": {
     "application/vnd.databricks.v1+cell": {
-     "cellMetadata": {},
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
      "inputWidgets": {},
      "nuid": "cfebdfdf-b87c-4a77-b97c-4697566a55fa",
      "showTitle": false,
@@ -255,6 +268,29 @@
   {
    "cell_type": "code",
    "execution_count": null,
+   "metadata": {
+    "application/vnd.databricks.v1+cell": {
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
+     "inputWidgets": {},
+     "nuid": "0d1f2e9e-db40-41fd-a6b9-bb4757db08b0",
+     "showTitle": false,
+     "title": ""
+    }
+   },
+   "outputs": [],
+   "source": [
+    "# Make sure you have write access to the ``home`` directory\n",
+    "home = os.path.join('/local_disk0', 'ift')\n",
+    "os.makedirs(home, exist_ok=True)\n",
+    "os.chdir(home)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 0,
    "metadata": {
     "application/vnd.databricks.v1+cell": {
      "cellMetadata": {
@@ -271,22 +307,26 @@
    "source": [
     "FT_API_args = Namespace(\n",
     "    model= 'mosaicml/mpt-7b', # Other examples: 'EleutherAI/gpt-neox-20b',\n",
-    "    train_data_path= 'main.streaming.random_large_table', # Other examples: 'tatsu-lab/alpaca/train', # '/Volumes/main/mosaic_hackathon/managed-volume/IFT/train.jsonl'  # 'mosaicml/dolly_hhrlhf/train'\n",
+    "    train_data_path= 'mosaicml/dolly_hhrlhf/train',  # Other examples: '/path/to/train.jsonl', 'catalog.schema.table'\n",
     "    task_type='INSTRUCTION_FINETUNE',\n",
     "    training_duration=3,\n",
     "    context_length=2048,\n",
     ")\n",
     "\n",
-    "temporary_jsonl_data_path = '/Volumes/main/mosaic_hackathon/managed-volume/IFT/ft_data_11Jan24_3/train'\n",
-    "os.environ['HF_DATASETS_CACHE'] = '/tmp/'\n",
-    "os.makedirs(temporary_jsonl_data_path, exist_ok=True)"
+    "temporary_jsonl_data_path = os.path.join(home, 'ft_data_11Jan24_3/train')\n",
+    "os.environ['HF_DATASETS_CACHE'] = os.path.join(home, 'hf_cache')\n",
+    "os.makedirs(temporary_jsonl_data_path, exist_ok=True)\n",
+    "os.makedirs(os.environ['HF_DATASETS_CACHE'], exist_ok=True)"
    ]
   },
   {
    "cell_type": "markdown",
    "metadata": {
     "application/vnd.databricks.v1+cell": {
-     "cellMetadata": {},
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
      "inputWidgets": {},
      "nuid": "39c45005-1a77-4162-b9e4-bd8df6f5ec69",
      "showTitle": false,
@@ -362,7 +402,10 @@
    "cell_type": "markdown",
    "metadata": {
     "application/vnd.databricks.v1+cell": {
-     "cellMetadata": {},
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
      "inputWidgets": {},
      "nuid": "06d46367-bd32-473a-9f16-1b34a8dd9356",
      "showTitle": false,
@@ -377,7 +420,10 @@
    "cell_type": "markdown",
    "metadata": {
     "application/vnd.databricks.v1+cell": {
-     "cellMetadata": {},
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
      "inputWidgets": {},
      "nuid": "1a28320a-a2a1-4f3c-a0cd-ad6045a24f64",
      "showTitle": false,
@@ -467,7 +513,10 @@
    "cell_type": "markdown",
    "metadata": {
     "application/vnd.databricks.v1+cell": {
-     "cellMetadata": {},
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
      "inputWidgets": {},
      "nuid": "9713a0ce-80f4-4187-b10b-4223b17fe4c1",
      "showTitle": false,
@@ -506,7 +555,10 @@
    "cell_type": "markdown",
    "metadata": {
     "application/vnd.databricks.v1+cell": {
-     "cellMetadata": {},
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
      "inputWidgets": {},
      "nuid": "7249e9e6-1ea7-4fc9-8959-8a17d62a9fb4",
      "showTitle": false,
@@ -547,7 +599,10 @@
    "cell_type": "markdown",
    "metadata": {
     "application/vnd.databricks.v1+cell": {
-     "cellMetadata": {},
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
      "inputWidgets": {},
      "nuid": "6699f47f-9b53-47da-95c0-b862c5826d0a",
      "showTitle": false,
@@ -562,7 +617,10 @@
    "cell_type": "markdown",
    "metadata": {
     "application/vnd.databricks.v1+cell": {
-     "cellMetadata": {},
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
      "inputWidgets": {},
      "nuid": "dd37fdce-62d0-493e-bfa9-d823634b2a0d",
      "showTitle": false,
@@ -624,12 +682,66 @@
    "source": [
     "FT_API_args = Namespace(\n",
     "    model= 'mosaicml/mpt-7b',\n",
-    "    train_data_path= '/Volumes/main/mosaic_hackathon/managed-volume/ABT',\n",
+    "    train_data_path= os.path.join(home, 'ABT'), # this is the path to your collection of txt files\n",
     "    task_type='CONTINUED_PRETRAIN',\n",
     "    training_duration=3,\n",
-    "    context_length=2048,\n",
+    "    context_length=8,\n",
     ")\n",
-    "temporary_mds_output_path = '/Volumes/main/mosaic_hackathon/managed-volume/{your_username}/mds_data_11Jan24_5'"
+    "temporary_mds_output_path = os.path.join(home, 'mds_data_11Jan24_5')"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {
+    "application/vnd.databricks.v1+cell": {
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
+     "inputWidgets": {},
+     "nuid": "fc2e4e8b-7700-47c4-bb21-ae4c389f39a2",
+     "showTitle": false,
+     "title": ""
+    }
+   },
+   "source": [
+    "Generate a synthetic dataset. Replace train_data_path with your raw data path in practice."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 0,
+   "metadata": {
+    "application/vnd.databricks.v1+cell": {
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
+     "inputWidgets": {},
+     "nuid": "10f08422-5091-4e64-b3f7-54928584cd60",
+     "showTitle": false,
+     "title": ""
+    }
+   },
+   "outputs": [],
+   "source": [
+    "def generate_synthetic_dataset(folder_path, num_files=128):\n",
+    "    \"\"\"Generate a synthetic dataset of text files with random words.\"\"\"\n",
+    "    def generate_random_words(num_words=50):\n",
+    "        words = [\"apple\", \"banana\", \"cherry\", \"date\", \"elderberry\", \"fig\", \"grape\", \"honeydew\", \"kiwi\", \"lemon\", \"mango\", \"nectarine\", \"orange\", \"papaya\", \"quince\", \"raspberry\", \"strawberry\", \"tangerine\", \"ugli\", \"vanilla\", \"watermelon\", \"xigua\", \"yam\", \"zucchini\"]\n",
+    "        return ' '.join(random.choice(words) for _ in range(num_words))\n",
+    "\n",
+    "    if not os.path.exists(folder_path):\n",
+    "        os.makedirs(folder_path)\n",
+    "    \n",
+    "    for i in range(num_files):\n",
+    "        file_path = os.path.join(folder_path, f\"file_{i}.txt\")\n",
+    "        with open(file_path, 'w') as file:\n",
+    "            file.write(generate_random_words())\n",
+    "\n",
+    "    print(f\"Generated {num_files} files in '{folder_path}'.\")\n",
+    "\n",
+    "generate_synthetic_dataset(FT_API_args.train_data_path)"
    ]
   },
   {
@@ -656,7 +768,10 @@
    "cell_type": "markdown",
    "metadata": {
     "application/vnd.databricks.v1+cell": {
-     "cellMetadata": {},
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
      "inputWidgets": {},
      "nuid": "c21e7d1b-db34-4e5d-b6d9-190dc75170d3",
      "showTitle": false,
@@ -688,6 +803,27 @@
   {
    "cell_type": "code",
    "execution_count": null,
+   "metadata": {
+    "application/vnd.databricks.v1+cell": {
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
+     "inputWidgets": {},
+     "nuid": "f5aea2a8-db29-40c9-8ed2-b6a1d032e7ab",
+     "showTitle": false,
+     "title": ""
+    }
+   },
+   "outputs": [],
+   "source": [
+    "import os\n",
+    "os.makedirs(temporary_mds_output_path, exist_ok=True)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 0,
    "metadata": {
     "application/vnd.databricks.v1+cell": {
      "cellMetadata": {
@@ -734,7 +870,10 @@
    "cell_type": "markdown",
    "metadata": {
     "application/vnd.databricks.v1+cell": {
-     "cellMetadata": {},
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
      "inputWidgets": {},
      "nuid": "298eb990-9160-4e1b-958f-33dd2c11b54b",
      "showTitle": false,
@@ -776,7 +915,10 @@
    "execution_count": null,
    "metadata": {
     "application/vnd.databricks.v1+cell": {
-     "cellMetadata": {},
+     "cellMetadata": {
+      "byteLimit": 2048000,
+      "rowLimit": 10000
+     },
      "inputWidgets": {},
      "nuid": "e123669c-2f77-4d66-93eb-04efd546f39f",
      "showTitle": false,