wehub-resource-sync
199 lines
6.4 KiB
Plaintext
199 lines
6.4 KiB
Plaintext
{
|
|
"cells": [
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "6sr3Gk8hNDaF"
|
|
},
|
|
"source": []
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {
|
|
"id": "eQk5Dv3jyheg"
|
|
},
|
|
"source": [
|
|
"# If you are using Colab for free, we highly recommend you activate the T4 GPU\n",
|
|
"# hardware accelerator. Our models are designed to run with at least 16GB\n",
|
|
"# of RAM, activating T4 will grant the notebook 16GB of GDDR6 RAM as opposed\n",
|
|
"# to the ~13GB Colab gives automatically.\n",
|
|
"# To activate T4:\n",
|
|
"# 1. click on the \"Runtime\" tab\n",
|
|
"# 2. click on \"Change runtime type\"\n",
|
|
"# 3. select T4 GPU under Hardware Accelerator\n",
|
|
"# NOTE: there is a weekly usage limit on using T4 for free"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "td7dVt4fD_j6"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"!pip install llmware"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "IY8ysw9FElZR"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"!pip install --upgrade huggingface_hub\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 8,
|
|
"metadata": {
|
|
"id": "AeR-LvVXD27i"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"\n",
|
|
"\"\"\" This example shows a complex multi-part research analysis. In this example, we will:\n",
|
|
"\n",
|
|
" 1. Build a \"research\" library.\n",
|
|
" 2. Query the research library to identify topics of interest.\n",
|
|
" 3. Create an agent with several analytical tools: sentiment, emotions, topic, entities analysis\n",
|
|
" 4. Pass the results of our query to the agent to conduct multifaceted analysis.\n",
|
|
" 5. Apply a top-level filter ('sentiment') on the results from the query\n",
|
|
" 6. For any of the passages with negative sentiment, we will run a follow-up set of analyses.\n",
|
|
" 7. Finally, we will assemble the follow-up analysis into a list of detailed reports.\n",
|
|
"\"\"\"\n",
|
|
"\n",
|
|
"import os\n",
|
|
"import shutil\n",
|
|
"\n",
|
|
"from llmware.agents import LLMfx\n",
|
|
"from llmware.library import Library\n",
|
|
"from llmware.retrieval import Query\n",
|
|
"from llmware.configs import LLMWareConfig\n",
|
|
"from llmware.setup import Setup\n",
|
|
"\n",
|
|
"\n",
|
|
"\n",
|
|
"\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": 9,
|
|
"metadata": {
|
|
"id": "qG3lAj2nM372"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"def multistep_analysis():\n",
|
|
"\n",
|
|
" \"\"\" In this example, our objective is to research Microsoft history and rivalry in the 1980s with IBM. \"\"\"\n",
|
|
"\n",
|
|
" # step 1 - assemble source documents and create library\n",
|
|
"\n",
|
|
" print(\"update: Starting example - agent-multistep-analysis\")\n",
|
|
"\n",
|
|
" # note: lines 38-49 attempt to automatically pull sample document into local path\n",
|
|
" # depending upon permissions in your environment, you may need to set up directly\n",
|
|
" # if you pull down the samples files with Setup().load_sample_files(), in the Books folder,\n",
|
|
" # you will find the source: \"Bill-Gates-Biography.pdf\"\n",
|
|
" # if you have pulled sample documents in the past, then to update to latest: set over_write=True\n",
|
|
"\n",
|
|
" print(\"update: Loading sample files\")\n",
|
|
"\n",
|
|
" sample_files_path = Setup().load_sample_files(over_write=False)\n",
|
|
" bill_gates_bio = \"Bill-Gates-Biography.pdf\"\n",
|
|
" path_to_bill_gates_bio = os.path.join(sample_files_path, \"Books\", bill_gates_bio)\n",
|
|
"\n",
|
|
" microsoft_folder = os.path.join(LLMWareConfig().get_tmp_path(), \"example_microsoft\")\n",
|
|
"\n",
|
|
" print(\"update: attempting to create source input folder at path: \", microsoft_folder)\n",
|
|
"\n",
|
|
" if not os.path.exists(microsoft_folder):\n",
|
|
" os.mkdir(microsoft_folder)\n",
|
|
" os.chmod(microsoft_folder, 0o777)\n",
|
|
" shutil.copy(path_to_bill_gates_bio,os.path.join(microsoft_folder, bill_gates_bio))\n",
|
|
"\n",
|
|
" # create library\n",
|
|
" print(\"update: creating library and parsing source document\")\n",
|
|
"\n",
|
|
" LLMWareConfig().set_active_db(\"sqlite\")\n",
|
|
" my_lib = Library().create_new_library(\"microsoft_history_0210_1\")\n",
|
|
" my_lib.add_files(microsoft_folder)\n",
|
|
"\n",
|
|
" # run our first query - \"ibm\"\n",
|
|
" query = \"ibm\"\n",
|
|
" search_results = Query(my_lib).text_query(query)\n",
|
|
" print(f\"update: executing query to filter to key passages - {query} - results found - {len(search_results)}\")\n",
|
|
"\n",
|
|
" # create an agent and load several tools that we will be using\n",
|
|
" agent = LLMfx()\n",
|
|
" agent.load_tool_list([\"sentiment\", \"emotions\", \"topic\", \"tags\", \"ner\", \"answer\"])\n",
|
|
"\n",
|
|
" # load the search results into the agent's work queue\n",
|
|
" agent.load_work(search_results)\n",
|
|
"\n",
|
|
" while True:\n",
|
|
"\n",
|
|
" agent.sentiment()\n",
|
|
"\n",
|
|
" if not agent.increment_work_iteration():\n",
|
|
" break\n",
|
|
"\n",
|
|
" # analyze sections where the sentiment on ibm was negative\n",
|
|
" follow_up_list = agent.follow_up_list(key=\"sentiment\", value=\"negative\")\n",
|
|
"\n",
|
|
" for job_index in follow_up_list:\n",
|
|
"\n",
|
|
" # follow-up 'deep dive' on selected text that references ibm negatively\n",
|
|
" agent.set_work_iteration(job_index)\n",
|
|
" agent.exec_multitool_function_call([\"tags\", \"emotions\", \"topics\", \"ner\"])\n",
|
|
" agent.answer(\"What is a brief summary?\", key=\"summary\")\n",
|
|
"\n",
|
|
" my_report = agent.show_report(follow_up_list)\n",
|
|
"\n",
|
|
" activity_summary = agent.activity_summary()\n",
|
|
"\n",
|
|
" for entries in my_report:\n",
|
|
" print(\"my report entries: \", entries)\n",
|
|
"\n",
|
|
" return my_report"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {
|
|
"id": "NDKDVuEoM9rL"
|
|
},
|
|
"outputs": [],
|
|
"source": [
|
|
"\n",
|
|
"if __name__ == \"__main__\":\n",
|
|
"\n",
|
|
" multistep_analysis()\n",
|
|
"\n"
|
|
]
|
|
}
|
|
],
|
|
"metadata": {
|
|
"accelerator": "GPU",
|
|
"colab": {
|
|
"gpuType": "T4",
|
|
"provenance": []
|
|
},
|
|
"kernelspec": {
|
|
"display_name": "Python 3",
|
|
"name": "python3"
|
|
},
|
|
"language_info": {
|
|
"name": "python"
|
|
}
|
|
},
|
|
"nbformat": 4,
|
|
"nbformat_minor": 0
|
|
}
|