Skip to content

Commit

Permalink
[RAG]update rag with text examples (#1967)
Browse files Browse the repository at this point in the history 
including:
1. update rag with text examples
2. change the PDF loader for better accuracy
  • Loading branch information
@openvino-dev-samples
openvino-dev-samples authored Apr 26, 2024
1 parent 631607a commit 1283dc0
Showing 1 changed file with 95 additions and 41 deletions.
136 changes: 95 additions & 41 deletions notebooks/llm-rag-langchain/llm-rag-langchain.ipynb
View file
Original file line number Diff line number Diff line change
Expand Up @@ -59,10 +59,21 @@
},
{
"cell_type": "code",
"execution_count": 1,
"execution_count": 50,
"id": "1f077b32-5d36-44b0-9041-407e996283a3",
"metadata": {},
"outputs": [],
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"\u001b[33mWARNING: Skipping openvino-dev as it is not installed.\u001b[0m\u001b[33m\n",
"\u001b[0m\u001b[33mWARNING: Skipping openvino as it is not installed.\u001b[0m\u001b[33m\n",
"\u001b[0mNote: you may need to restart the kernel to use updated packages.\n",
"Note: you may need to restart the kernel to use updated packages.\n"
]
}
],
"source": [
"%pip uninstall -q -y openvino-dev openvino openvino-nightly optimum optimum-intel\n",
"%pip install -q --extra-index-url https://download.pytorch.org/whl/cpu\\\n",
Expand All @@ -72,7 +83,7 @@
"\"accelerate\"\\\n",
"\"openvino-nightly\"\\\n",
"\"gradio\"\\\n",
"\"onnx\" \"einops\" \"transformers_stream_generator\" \"tiktoken\" \"transformers>=4.38.1\" \"bitsandbytes\" \"chromadb\" \"sentence_transformers\" \"langchain>=0.1.15\" \"langchainhub\" \"unstructured\" \"scikit-learn\" \"python-docx\" \"pdfminer.six\" "
"\"onnx\" \"einops\" \"transformers_stream_generator\" \"tiktoken\" \"transformers>=4.38.1\" \"bitsandbytes\" \"chromadb\" \"sentence_transformers\" \"langchain>=0.1.15\" \"langchainhub\" \"unstructured\" \"scikit-learn\" \"python-docx\" \"pypdf\" "
]
},
{
Expand All @@ -85,19 +96,36 @@
"import shutil\n",
"from pathlib import Path\n",
"import requests\n",
"import io\n",
"\n",
"# fetch model configuration\n",
"\n",
"config_shared_path = Path(\"../../utils/llm_config.py\")\n",
"config_dst_path = Path(\"llm_config.py\")\n",
"text_example_en_path = Path(\"text_example_en.pdf\")\n",
"text_example_cn_path = Path(\"text_example_cn.pdf\")\n",
"text_example_en = \"https://github.com/openvinotoolkit/openvino_notebooks/files/15039728/Platform.Brief_Intel.vPro.with.Intel.Core.Ultra_Final.pdf\"\n",
"text_example_cn = \"https://github.com/openvinotoolkit/openvino_notebooks/files/15039713/Platform.Brief_Intel.vPro.with.Intel.Core.Ultra_Final_CH.pdf\"\n",
"\n",
"if not config_dst_path.exists():\n",
" if config_shared_path.exists():\n",
" shutil.copy(config_shared_path, config_dst_path)\n",
" else:\n",
" r = requests.get(url=\"https://raw.githubusercontent.com/openvinotoolkit/openvino_notebooks/latest/utils/llm_config.py\")\n",
" with open(\"llm_config.py\", \"w\") as f:\n",
" f.write(r.text)"
" f.write(r.text)\n",
"\n",
"if not text_example_en_path.exists():\n",
" r = requests.get(url=text_example_en)\n",
" content = io.BytesIO(r.content)\n",
" with open(\"text_example_en.pdf\", \"wb\") as f:\n",
" f.write(content.read())\n",
"\n",
"if not text_example_cn_path.exists():\n",
" r = requests.get(url=text_example_cn)\n",
" content = io.BytesIO(r.content)\n",
" with open(\"text_example_cn.pdf\", \"wb\") as f:\n",
" f.write(content.read())"
]
},
{
Expand Down Expand Up @@ -173,7 +201,7 @@
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "8aff9565ac6f4c0993359ef2df59734c",
"model_id": "ed92457ad3cf42b2a22ca224141fe738",
"version_major": 2,
"version_minor": 0
},
Expand Down Expand Up @@ -214,12 +242,12 @@
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "84cb9d80223e4cab856c3e5ea9d88ea0",
"model_id": "aad6acd73a9a40fbae5ec40195512a80",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
"Dropdown(description='Model:', index=10, options=('tiny-llama-1b-chat', 'gemma-2b-it', 'red-pajama-3b-chat', '…"
"Dropdown(description='Model:', index=9, options=('tiny-llama-1b-chat', 'gemma-2b-it', 'red-pajama-3b-chat', 'g…"
]
},
"execution_count": 4,
Expand Down Expand Up @@ -250,7 +278,7 @@
"name": "stdout",
"output_type": "stream",
"text": [
"Selected LLM model llama-3-8b-instruct\n"
"Selected LLM model neural-chat-7b-v3-1\n"
]
}
],
Expand Down Expand Up @@ -308,7 +336,7 @@
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "82f25885896749728c96d9f909f65a5a",
"model_id": "009a1f80d9e24964a084d5d31cfecccd",
"version_major": 2,
"version_minor": 0
},
Expand All @@ -322,7 +350,7 @@
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "1731c06ddd1b4ad792516efb23b51f80",
"model_id": "f79b9054accd4501acad346d05ef3c6d",
"version_major": 2,
"version_minor": 0
},
Expand All @@ -336,7 +364,7 @@
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "6a8bd8732be14625ba4474989c4d6545",
"model_id": "a1af239531b44c04a2cac15dacb629bd",
"version_major": 2,
"version_minor": 0
},
Expand Down Expand Up @@ -522,7 +550,7 @@
"name": "stdout",
"output_type": "stream",
"text": [
"Size of model with INT4 compressed weights is 5106.56 MB\n"
"Size of model with INT4 compressed weights is 5069.90 MB\n"
]
}
],
Expand Down Expand Up @@ -562,7 +590,7 @@
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "7cdeb3e1ce7c464f84c81d118c3856b8",
"model_id": "7ff9bbb6909d40968b638df90683f360",
"version_major": 2,
"version_minor": 0
},
Expand Down Expand Up @@ -648,7 +676,7 @@
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "060d8facb29e4ab99ba5c5a51a36a0cd",
"model_id": "ff064eaa037440a7b22af61047422e9c",
"version_major": 2,
"version_minor": 0
},
Expand Down Expand Up @@ -741,12 +769,12 @@
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "31871e2ec744496ca8ccfad377310c5c",
"model_id": "1448cdadd58b41fa8116605d114285ef",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
"Dropdown(description='Device:', options=('CPU', 'GPU.0', 'GPU.1', 'AUTO'), value='CPU')"
"Dropdown(description='Device:', options=('CPU', 'GPU', 'AUTO'), value='CPU')"
]
},
"execution_count": 15,
Expand Down Expand Up @@ -803,12 +831,12 @@
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "1c269bd0ac424e8eb1a9c5738d1709d6",
"model_id": "321d8385fc0345498f07154637c09eb4",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
"Dropdown(description='Device:', options=('CPU', 'GPU.0', 'GPU.1', 'AUTO'), value='CPU')"
"Dropdown(description='Device:', options=('CPU', 'GPU', 'AUTO'), value='CPU')"
]
},
"execution_count": 17,
Expand Down Expand Up @@ -865,12 +893,12 @@
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "889f3eb31f0d4cfaacefb453d58065df",
"model_id": "1f833b438e18488c9b9c85cb87786b26",
"version_major": 2,
"version_minor": 0
},
"text/plain": [
"Dropdown(description='Device:', options=('CPU', 'GPU.0', 'GPU.1', 'AUTO'), value='CPU')"
"Dropdown(description='Device:', options=('CPU', 'GPU', 'AUTO'), value='CPU')"
]
},
"execution_count": 19,
Expand Down Expand Up @@ -941,16 +969,14 @@
"name": "stderr",
"output_type": "stream",
"text": [
"2024-04-19 00:18:35.178975: I tensorflow/core/util/port.cc:111] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.\n",
"2024-04-19 00:18:35.181441: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers on your machine, GPU will not be used.\n",
"2024-04-19 00:18:35.212245: E tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:9342] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered\n",
"2024-04-19 00:18:35.212267: E tensorflow/compiler/xla/stream_executor/cuda/cuda_fft.cc:609] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered\n",
"2024-04-19 00:18:35.212295: E tensorflow/compiler/xla/stream_executor/cuda/cuda_blas.cc:1518] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered\n",
"2024-04-19 00:18:35.219017: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers on your machine, GPU will not be used.\n",
"2024-04-19 00:18:35.219908: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n",
"2024-04-25 23:24:53.325545: I tensorflow/core/util/port.cc:111] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.\n",
"2024-04-25 23:24:53.329247: I tensorflow/tsl/cuda/cudart_stub.cc:28] Could not find cuda drivers on your machine, GPU will not be used.\n",
"2024-04-25 23:24:53.374367: E tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:9342] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered\n",
"2024-04-25 23:24:53.374399: E tensorflow/compiler/xla/stream_executor/cuda/cuda_fft.cc:609] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered\n",
"2024-04-25 23:24:53.374437: E tensorflow/compiler/xla/stream_executor/cuda/cuda_blas.cc:1518] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered\n",
"2024-04-25 23:24:53.385838: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.\n",
"To enable the following instructions: AVX2 AVX512F AVX512_VNNI FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
"2024-04-19 00:18:36.155999: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n",
"The installed version of bitsandbytes was compiled without GPU support. 8-bit optimizers, 8-bit multiplication, and GPU quantization are unavailable.\n",
"2024-04-25 23:24:54.133894: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT\n",
"Compiling the model to CPU ...\n"
]
},
Expand Down Expand Up @@ -1050,7 +1076,7 @@
{
"data": {
"application/vnd.jupyter.widget-view+json": {
"model_id": "b815fe3af56542cbb08c2d06e9e9dd26",
"model_id": "42271e02cc19435f89659d127241fb1a",
"version_major": 2,
"version_minor": 0
},
Expand Down Expand Up @@ -1096,20 +1122,26 @@
"id": "f7f708db-8de1-4efd-94b2-fcabc48d52f4",
"metadata": {},
"outputs": [
{
"name": "stderr",
"output_type": "stream",
"text": [
"The argument `trust_remote_code` is to be used along with export=True. It will be ignored.\n"
]
},
{
"name": "stdout",
"output_type": "stream",
"text": [
"Loading model from llama-3-8b-instruct/INT4_compressed_weights\n"
"Loading model from neural-chat-7b-v3-1/INT4_compressed_weights\n"
]
},
{
"name": "stderr",
"output_type": "stream",
"text": [
"Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.\n",
"The argument `trust_remote_code` is to be used along with export=True. It will be ignored.\n",
"Compiling the model to CPU ...\n"
"Compiling the model to CPU ...\n",
"Setting `pad_token_id` to `eos_token_id`:2 for open-end generation.\n"
]
},
{
Expand Down Expand Up @@ -1208,7 +1240,7 @@
"from langchain.document_loaders import (\n",
" CSVLoader,\n",
" EverNoteLoader,\n",
" PDFMinerLoader,\n",
" PyPDFLoader,\n",
" TextLoader,\n",
" UnstructuredEPubLoader,\n",
" UnstructuredHTMLLoader,\n",
Expand Down Expand Up @@ -1255,11 +1287,30 @@
" \".html\": (UnstructuredHTMLLoader, {}),\n",
" \".md\": (UnstructuredMarkdownLoader, {}),\n",
" \".odt\": (UnstructuredODTLoader, {}),\n",
" \".pdf\": (PDFMinerLoader, {}),\n",
" \".pdf\": (PyPDFLoader, {}),\n",
" \".ppt\": (UnstructuredPowerPointLoader, {}),\n",
" \".pptx\": (UnstructuredPowerPointLoader, {}),\n",
" \".txt\": (TextLoader, {\"encoding\": \"utf8\"}),\n",
"}"
"}\n",
"\n",
"chinese_examples = [\n",
" [\"英特尔®酷睿™ Ultra处理器可以降低多少功耗？\"],\n",
" [\"相比英特尔之前的移动处理器产品，英特尔®酷睿™ Ultra处理器的AI推理性能提升了多少？\"],\n",
" [\"英特尔博锐® Enterprise系统提供哪些功能？\"],\n",
"]\n",
"\n",
"english_examples = [\n",
" [\"How much power consumption can Intel® Core™ Ultra Processors help save?\"],\n",
" [\"Compared to Intel’s previous mobile processor, what is the advantage of Intel® Core™ Ultra Processors for Artificial Intelligence?\"],\n",
" [\"What can Intel vPro® Enterprise systems offer?\"],\n",
"]\n",
"\n",
"if model_language.value == \"English\":\n",
" text_example_path = \"text_example_en.pdf\"\n",
"else:\n",
" text_example_path = \"text_example_cn.pdf\"\n",
"\n",
"examples = chinese_examples if (model_language.value == \"Chinese\") else english_examples"
]
},
{
Expand Down Expand Up @@ -1388,6 +1439,7 @@
"\n",
" global combine_docs_chain\n",
" combine_docs_chain = create_stuff_documents_chain(llm, prompt)\n",
"\n",
" global rag_chain\n",
" rag_chain = create_retrieval_chain(retriever, combine_docs_chain)\n",
"\n",
Expand Down Expand Up @@ -1488,6 +1540,7 @@
" with gr.Column(scale=1):\n",
" docs = gr.File(\n",
" label=\"Step 1: Load text files\",\n",
" value=[text_example_path],\n",
" file_count=\"multiple\",\n",
" file_types=[\n",
" \".csv\",\n",
Expand Down Expand Up @@ -1608,6 +1661,7 @@
" submit = gr.Button(\"Submit\")\n",
" stop = gr.Button(\"Stop\")\n",
" clear = gr.Button(\"Clear\")\n",
" gr.Examples(examples, inputs=msg, label=\"Click on any example and press the 'Submit' button\")\n",
" retriever_argument = gr.Accordion(\"Retriever Configuration\", open=True)\n",
" with retriever_argument:\n",
" with gr.Row():\n",
Expand Down Expand Up @@ -1685,11 +1739,11 @@
" queue=False,\n",
" )\n",
" clear.click(lambda: None, None, chatbot, queue=False)\n",
" vector_search_top_k.change(\n",
" vector_search_top_k.release(\n",
" update_retriever,\n",
" [vector_search_top_k, vector_rerank_top_n, do_rerank, search_method],\n",
" )\n",
" vector_rerank_top_n.change(\n",
" vector_rerank_top_n.release(\n",
" update_retriever,\n",
" [vector_search_top_k, vector_rerank_top_n, do_rerank, search_method],\n",
" )\n",
Expand Down Expand Up @@ -1726,7 +1780,7 @@
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"display_name": "Python 3 (ipykernel)",
"language": "python",
"name": "python3"
},
Expand All @@ -1740,7 +1794,7 @@
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.11.4"
"version": "3.10.12"
},
"openvino_notebooks": {
"imageUrl": "https://github.com/openvinotoolkit/openvino_notebooks/assets/29454499/304aa048-f10c-41c6-bb31-6d2bfdf49cf5",
Expand Down

0 comments on commit 1283dc0

Please sign in to comment.