Fine-Tuning Open-Source LLMs (Llama, Mistral, Qwen)
Professional skills preparation with worked examples, written practice and independent learning assessments.
Professional Fine-Tuning Open-Source LLMs (Llama, Mistral, Qwen) course covering: Open-weight models landscape and licences; Running models locally with Ollama and Hugging Face; Datasets for instruction tuning; Full fine-tuning versus LoRA and QLoRA; Training with Hugging Face TRL and Unsloth; Preference tuning: RLHF, DPO and evaluation; Quantization, serving with vLLM and cost; Capstone: fine-tune and deploy a domain model. Practical lessons, hands-on labs, exercises, quizzes and a final exam. Includes lessons, supplementary YouTube tutorials, practical exercises, quizzes and free/paid practice tests. Erudex is independent, not affiliated with the source institution or awarding body. Its 80% learning checks are not official grades or university credit. Content requires qualified subject review before use as a complete teaching programme.
Curriculum
One complete track with its own final exam and certificate.
Professional Fine-Tuning Open-Source LLMs (Llama, Mistral, Qwen) course covering: Open-weight models landscape and licences; Running models locally with Ollama and Hugging Face; Datasets for instruction tuning; Full fine-tuning versus LoRA and QLoRA; Training with Hugging Face TRL and Unsloth; Preference tuning: RLHF, DPO and evaluation; Quantization, serving with vLLM and cost; Capstone: fine-tune and deploy a domain model. Practical lessons, hands-on labs, exercises, quizzes and a final exam.
Open-weight models landscape and licences
- 1. Mapping the open-weight LLM landscape Lab60 min
- 2. Reading licences and translating terms into obligations Lab60 min
- 3. Matching checkpoints to tasks, hardware, and risk Lab60 min
- 4. Creating an auditable model adoption and release plan Lab60 min