diff --git a/examples/partners/model_selection_guide/model_selection_guide.ipynb b/examples/partners/model_selection_guide/model_selection_guide.ipynb index 859f1e5fa4..88755463f5 100644 --- a/examples/partners/model_selection_guide/model_selection_guide.ipynb +++ b/examples/partners/model_selection_guide/model_selection_guide.ipynb @@ -61,7 +61,9 @@ "\n", "### OpenAI Model Evolution \n", "\n", - "![OpenAI Model Evolution](./images/2.2_model_evolution.png)\n", + + "![OpenAI Model Evolution](../../../images/2.2_model_evolution.png)\n", + "\n", "### Key Characteristics\n", "\n", @@ -79,7 +81,9 @@ "\n", "## 3A. Use Case: Long-Context RAG for Legal Q&A\n", "\n", - "![Long-Context RAG for Legal Q&A](images/3A_rag_task_card.png)\n", + + "![Long-Context RAG for Legal Q&A](../../../images/3A_rag_task_card.png)\n", + "## 🗂️ TL;DR Matrix\n", "\n", "This table summarizes the core technology choices and their rationale for **this specific Long-Context Agentic RAG implementation**.\n", @@ -133,7 +137,9 @@ "id": "db9bad1b", "metadata": {}, "source": [ - "![Hierarchical Router](images/3A_rag_hierarchical_router.png)\n", + + "![Hierarchical Router](../../../images/3A_rag_hierarchical_router.png)\n", + "\n", "\n", "## Agentic RAG System: Model Usage\n", @@ -1815,7 +1821,9 @@ "================================================================================\n", "\n", "## 3B. Use Case: AI Co-Scientist for Pharma R&D\n", - "![AI Co-Scientist for Pharma R&D](images/3B_reasoning_task_card.png)\n", + + "![AI Co-Scientist for Pharma R&D](../../../images/3B_reasoning_task_card.png)\n", + "\n", "This section details how to build an AI system that functions as a \"co-scientist\" to accelerate experimental design in pharmaceutical R&D, focusing on optimizing a drug synthesis process under specific constraints.\n", "\n", @@ -1855,7 +1863,9 @@ "\n", "The system employs a multi-agent architecture that emulates a high-performing scientific team. Different AI components, acting in specialized roles (such as ideation, critique, and learning from outcomes), collaborate using various models and tools to execute the workflow.\n", "\n", - "![AI Co-Scientist Architecture](images/3B_coscientist_architecture.png)\n", + + "![AI Co-Scientist Architecture](../../../images/3B_coscientist_architecture.png)\n", + "\n", "### 2.1. **Scientist Input & Constraints:** \n", "The process starts with the scientist defining the goal, target compound, and constraints." @@ -2463,7 +2473,9 @@ "\n", "## 3C. Use Case: Insurance Claim Processing\n", "\n", - "![](./images/3C_insurance_task_card.png)\n", + + "![](../../../images/3C_insurance_task_card.png)\n", + "\n", "Many businesses are faced with the task of digitizing hand filled forms. In this section, we will demonstrate how OpenAI can be used to digitize and validate a hand filled insurance form. While this is a common problem for insurance, the same techniques can be applied to a variety of other industries and forms, for example tax forms, invoices, and more.\n", "\n", @@ -2493,7 +2505,9 @@ "\n", "The high level basic architecture of the solution is shown below.\n", "\n", - "![](./images/3C_insurance_architecture.png)\n", + + "![](../../../images/3C_insurance_architecture.png)\n", + "\n", "This task is complex and requires a wide variety of model capabilities, including vision, function calling, reasoning, and structured output. While `o3` is capable of doing all of these at once, we found during experimentation that `o4-mini` alone was not sufficient to achieve the necessary performance. Due to the higher relative costs of `o3`, we instead opted for a two-stage approach.\n", "\n", @@ -2503,7 +2517,9 @@ "\n", "To demonstrate concretely how this works, let's look at a sample image of an insurance form.\n", "\n", - "![](./images/3C_insurance_form.png)\n", + + "![](../../../images/3C_insurance_form.png)\n", + "\n", "While the form itself is fairly straightforward, there is missing data and ambiguous information that will be difficult for a traditional OCR system to fill out correctly. First, notice that the zip code and county have been omitted. Second, the email address of the user is ambiguous \\-- it could be `jsmith1@gmail.com` or `jsmithl@gmail.com`. In the following sections, we will walk through how a well-designed solution can handle these ambiguities and return the correct form results.\n", "\n", @@ -3186,7 +3202,10 @@ "\n", "## Adaptation Decision Tree\n", "\n", - "![Model Selection Decision Tree](images/3D_model_selection_flowchart.png)\n", + + "![Model Selection Decision Tree](../../../images/3D_model_selection_flowchart.png)\n", + + "\n", "## Communicating Model Selection to Non-Technical Stakeholders\n", "\n", @@ -3286,10 +3305,12 @@ "\n", "## Contributors\n", "\n", - "- Kashyap Coimbatore Murali\n", - "- Nate Harada \n", - "- Sai Prashanth Soundararaj \n", - "- Shikhar Kwatra " + + "- [Kashyap Coimbatore Murali](https://www.linkedin.com/in/kashyap-murali/)\n", + "- [Nate Harada](https://www.linkedin.com/in/nate-harada/) \n", + "- [Sai Prashanth Soundararaj](https://www.linkedin.com/in/saiprashanths/)\n", + "- [Shikhar Kwatra](https://www.linkedin.com/in/shikharkwatra/)" + ] } ], diff --git a/images/2.2_model_evolution.png b/images/2.2_model_evolution.png new file mode 100644 index 0000000000..87f09ef350 Binary files /dev/null and b/images/2.2_model_evolution.png differ diff --git a/images/3A_rag_hierarchical_router.png b/images/3A_rag_hierarchical_router.png new file mode 100644 index 0000000000..0d1926b7c9 Binary files /dev/null and b/images/3A_rag_hierarchical_router.png differ diff --git a/images/3A_rag_task_card.png b/images/3A_rag_task_card.png new file mode 100644 index 0000000000..7761c85786 Binary files /dev/null and b/images/3A_rag_task_card.png differ diff --git a/images/3B_coscientist_architecture.png b/images/3B_coscientist_architecture.png new file mode 100644 index 0000000000..cea31baef5 Binary files /dev/null and b/images/3B_coscientist_architecture.png differ diff --git a/images/3B_reasoning_task_card.png b/images/3B_reasoning_task_card.png new file mode 100644 index 0000000000..c744242332 Binary files /dev/null and b/images/3B_reasoning_task_card.png differ diff --git a/images/3C_insurance_architecture.png b/images/3C_insurance_architecture.png new file mode 100644 index 0000000000..74342bd3c8 Binary files /dev/null and b/images/3C_insurance_architecture.png differ diff --git a/images/3C_insurance_form.png b/images/3C_insurance_form.png new file mode 100644 index 0000000000..97dae9e95f Binary files /dev/null and b/images/3C_insurance_form.png differ diff --git a/images/3C_insurance_task_card.png b/images/3C_insurance_task_card.png new file mode 100644 index 0000000000..65a8157bf0 Binary files /dev/null and b/images/3C_insurance_task_card.png differ diff --git a/images/3D_model_selection_flowchart.png b/images/3D_model_selection_flowchart.png new file mode 100644 index 0000000000..59e2d6a72a Binary files /dev/null and b/images/3D_model_selection_flowchart.png differ