QuanLLM-v0.1-qm is the first experimental field branch of the QuanLLM main product (an AI expert for quantum physics research and teaching), focused on university-level quantum mechanics final exam preparation. Built on the MiniMind architecture, it targets concept explanation, formula derivation, and exercise solving.
General-purpose LLMs can answer some quantum mechanics questions, but they often fall short in formula rigor, textbook consistency, and exam focus. v0.1-qm attempts to deploy a more focused model using domain-specific data.
Targets high-frequency final exam topics: concept explanation, formula derivation, calculation problems, and proof questions, with structured responses.
Can be fully deployed and run on a Windows PC with an NVIDIA RTX 5090. Data never leaves your machine.
Supports automatic extraction from PDF, Word, and TXT textbooks, as well as ready-made QA datasets such as StackExchange.
Built on MiniMind, with open deployment and data-processing scripts that can be extended to other quantum physics branches.
Explains core concepts such as wave functions, the uncertainty principle, representation transformations, and identical particles in clear language to build intuition.
Step-by-step derivations of the Schrödinger equation, ladder operators, angular momentum commutation relations, and more.
Provides problem-solving strategies and step-by-step solutions for harmonic oscillators, hydrogen atoms, perturbation theory, and other typical exam topics.
A Streamlit-based chat interface with adjustable temperature, Top-p, and history length, supporting streaming output.
One-click scripts for data generation, continued pre-training, SFT, and inference. Supports the RTX 5090 Blackwell architecture.
Put your quantum mechanics textbook (PDF / Word / TXT) into the raw_books/ directory.
Automatically extract pre-training corpus, heuristically generate QA pairs, or build SFT data from existing datasets.
Continue pre-training + full-parameter SFT from MiniMind official weights, or run SFT-only.
Interact with the model via command line or the Streamlit WebUI to verify response quality.
The current demo is static. Full conversational functionality requires deploying the model locally and accessing it through the WebUI.
💡 Tip: Download the project and deploy the model locally to start real conversations. This page is for product demonstration only.
Download QuanLLM-v0.1-qm and deploy your own model using your textbooks or QA datasets.