From cff0c4e17455a0d1863356b112607df8eff51417 Mon Sep 17 00:00:00 2001 From: Amit Divekar <110670491+amitdevx@users.noreply.github.com> Date: Mon, 5 Jan 2026 14:47:33 +0000 Subject: [PATCH] Upgrade Gemini models from 2.5 to 3 - Update all model references from gemini-2.5 to gemini-3 - Upgrade classifier model to gemini-3-flash - Upgrade analyzer model to gemini-3-pro - Update documentation (ARCHITECTURE.md, README.md, FEATURES.md) - Update configuration (config.py) --- ARCHITECTURE.md | 10 +++++----- FEATURES.md | 6 +++--- README.md | 8 ++++---- profiler_agent/config.py | 4 ++-- 4 files changed, 14 insertions(+), 14 deletions(-) diff --git a/ARCHITECTURE.md b/ARCHITECTURE.md index a095bd9..33a15cf 100644 --- a/ARCHITECTURE.md +++ b/ARCHITECTURE.md @@ -56,7 +56,7 @@ User Query → Runner → Root Agent (professor_profiler_agent) ``` ## Technology Stack -- **LLM**: Google Gemini 2.5 (Flash for classification, Pro for analysis) +- **LLM**: Google Gemini 3 (Flash for classification, Pro for analysis) - **Framework**: Custom ADK (Agent Development Kit) - **PDF Processing**: pypdf - **Visualization**: matplotlib @@ -67,25 +67,25 @@ User Query → Runner → Root Agent (professor_profiler_agent) ## Agent Specifications ### Root Agent: professor_profiler_agent -- **Model**: gemini-2.5-pro +- **Model**: gemini-3-pro - **Role**: Orchestrator - **Tools**: read_pdf_content, visualize_trends, analyze_statistics - **Sub-agents**: taxonomist, trend_spotter, strategist ### Sub-Agent: taxonomist -- **Model**: gemini-2.5-flash (cost-effective for classification) +- **Model**: gemini-3-flash (cost-effective for classification) - **Role**: Question classification - **Output**: Tagged questions with topics and Bloom's levels - **Callback**: Suppresses intermediate output ### Sub-Agent: trend_spotter -- **Model**: gemini-2.5-pro (complex analysis) +- **Model**: gemini-3-pro (complex analysis) - **Role**: Statistical trend analysis - **Output**: Shift report with frequency and cognitive trends - **Callback**: Suppresses intermediate output ### Sub-Agent: strategist -- **Model**: gemini-2.5-pro +- **Model**: gemini-3-pro - **Role**: Study plan generation - **Output**: Hit List, Safe Zone, Drop List recommendations diff --git a/FEATURES.md b/FEATURES.md index 481f386..968865f 100644 --- a/FEATURES.md +++ b/FEATURES.md @@ -5,7 +5,7 @@ ### 1. Multi-Agent System ✓ - [x] **Hub-and-Spoke Architecture**: Root orchestrator (`professor_profiler_agent`) + 3 specialized sub-agents - [x] **Sequential Agents**: Taxonomist → Trend Spotter → Strategist workflow -- [x] **Agent Powered by LLM**: All agents use Google Gemini 2.5 (Flash/Pro) +- [x] **Agent Powered by LLM**: All agents use Google Gemini 3 (Flash/Pro) - [x] **Parallel Processing Ready**: Batch question classification infrastructure - [x] **Agent Delegation**: Root agent delegates specialized tasks @@ -98,7 +98,7 @@ - **Test Coverage**: 5 comprehensive tests - **Agent Count**: 4 (1 root + 3 sub-agents) - **Custom Tools**: 4 -- **API Integration**: Google Gemini 2.5 +- **API Integration**: Google Gemini 3 - **Architecture Pattern**: Hub-and-Spoke ## 🎯 Features Demonstrated @@ -156,7 +156,7 @@ python demo.py 2. **Tool Integration**: Custom tools with Gemini function calling 3. **State Management**: Sessions and long-term memory 4. **Production Patterns**: Logging, tracing, metrics, error handling -5. **API Integration**: Google Gemini 2.5 with streaming responses +5. **API Integration**: Google Gemini 3 with streaming responses 6. **Testing**: Integration tests and comprehensive validation 7. **Documentation**: Clear architecture and usage documentation diff --git a/README.md b/README.md index 52256ce..8e16f8e 100644 --- a/README.md +++ b/README.md @@ -1,7 +1,7 @@ # Professor Profiler [![Python](https://img.shields.io/badge/Python-3.11%2B-blue?style=flat&logo=python&logoColor=white)](https://www.python.org/) -[![Google Gemini](https://img.shields.io/badge/Google%20Gemini-2.5%20Pro-8E44AD?style=flat&logo=google&logoColor=white)](https://ai.google.dev/) +[![Google Gemini](https://img.shields.io/badge/Google%20Gemini-3%20Pro-8E44AD?style=flat&logo=google&logoColor=white)](https://ai.google.dev/) [![Architecture](https://img.shields.io/badge/Architecture-Hub%20%26%20Spoke-orange?style=flat)](https://github.com/uffamit/Professor_Profiler) [![License](https://img.shields.io/badge/License-MIT-green?style=flat)](LICENSE) [![Quality Assurance Pipeline](https://github.com/uffamit/Professor_Profiler/actions/workflows/quality-assurance.yml/badge.svg)](https://github.com/uffamit/Professor_Profiler/actions/workflows/quality-assurance.yml) @@ -13,7 +13,7 @@ ## Overview -**Professor Profiler** is not just a document reader; it is a **Hierarchical Multi-Agent System (HMAS)** designed to mimic the cognitive process of an expert tutor. By orchestrating specialized AI agents powered by **Google Gemini 2.5**, it ingests raw exam PDFs, breaks them down into cognitive components (Bloom's Taxonomy), identifies statistical patterns, and formulates actionable "Safe Zone" and "Hit List" study plans. +**Professor Profiler** is not just a document reader; it is a **Hierarchical Multi-Agent System (HMAS)** designed to mimic the cognitive process of an expert tutor. By orchestrating specialized AI agents powered by **Google Gemini 3**, it ingests raw exam PDFs, breaks them down into cognitive components (Bloom's Taxonomy), identifies statistical patterns, and formulates actionable "Safe Zone" and "Hit List" study plans. This project serves as a reference implementation for: * **Hub-and-Spoke Agent Architecture** @@ -43,7 +43,7 @@ flowchart TD end subgraph Agent_Layer [" Agent Hierarchy"] - Root[ROOT AGENT
Gemini 2.5 Pro
The Project Manager] + Root[ROOT AGENT
Gemini 3 Pro
The Project Manager] subgraph Workers ["Specialized Sub-Agents"] Taxonomist[Taxonomist
Gemini Flash
Topic & Bloom's Classification] @@ -124,7 +124,7 @@ sequenceDiagram | Component | Technology | Description | | :--- | :--- | :--- | | **Core Logic** | Python 3.10+ | Type-hinted, async-native codebase. | -| **LLM Engine** | Google Gemini 2.5 | Uses `Pro` for reasoning and `Flash` for high-speed tasks. | +| **LLM Engine** | Google Gemini 3 | Uses `Pro` for reasoning and `Flash` for high-speed tasks. | | **Orchestrator** | Google ADK (Custom) | Custom implementation of the Agent Development Kit patterns. | | **Document Processing** | `pypdf` | Robust text extraction from standardized exam papers. | | **Visualization** | `matplotlib` | Generates distribution bar charts and pie charts on the fly. | diff --git a/profiler_agent/config.py b/profiler_agent/config.py index becc0bd..ffbcabd 100644 --- a/profiler_agent/config.py +++ b/profiler_agent/config.py @@ -15,7 +15,7 @@ @dataclass class ProfilerConfiguration: - classifier_model: str = "gemini-2.5-flash" - analyzer_model: str = "gemini-2.5-pro" + classifier_model: str = "gemini-3-flash" + analyzer_model: str = "gemini-3-pro" config = ProfilerConfiguration()