> Markdown version of [/playlists/fine-tuning](https://www.wearedevelopers.com/playlists/fine-tuning). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md). --- # Playlist: Fine-tuning 19 videos · 22 moments · 49.6 minutes ## Cutting LLM Costs Without Cutting Quality: How to Beat Proprietary LLMs with Fine-Tuned Open Source - **Identifying when narrow tasks require custom model fine-tuning** (19:05, 1min) — Determining when to transition from prompt engineering and retrieval-augmented generation to supervised fine-tuning reduces latency and API depende... - **Trade-offs between proprietary, open weight, and fine-tuned models** (06:31, 3min) — Navigating model selection requires balancing organizational control, hosting responsibilities, and specific task complexities. [Learn more](https://www.wearedevelopers.com/videos/100151-cutting-llm-costs-without-cutting-quality-how-to-beat-proprietary-llms-with-fine-tuned-open-source) ## Fine-Tuning Small Language Models for Agentic AI - **Implementing simple supervised fine-tuning with the Unsloth library** (19:22, 1min) — Specific training prompts and the Unsloth library streamline the process of fine-tuning quantized models. - **Benchmark limitations and final recommendations for fine-tuning models** (25:38, 2min) — Despite evaluation limits, supervised fine-tuning offers clear return on investment for highly specialized sub-agent tasks. - **Methods for supervised fine-tuning and reinforcement learning** (12:46, 3min) — Data distillation enables supervised fine-tuning while GRPO offers reinforcement learning with verifiable rewards. [Learn more](https://www.wearedevelopers.com/videos/100352-fine-tuning-small-language-models-for-agentic-ai) ## Large Language Models ❤️ Knowledge Graphs - **Comparing fine-tuning to database retrieval and grounding** (03:14, 1min) — Grounding language models via accessible database context offers superior security and efficiency compared to full fine-tuning. [Learn more](https://www.wearedevelopers.com/videos/1154-large-language-models-knowledge-graphs) ## DevOps for AI: running LLMs in production with Kubernetes and KubeFlow - **Enhancing models with retrieval augmented generation and fine-tuning** (15:14, 3min) — Indexing proprietary documentation for retrieval prevents model hallucinations, while permanent fine-tuning adapts generic models to domain-specifi... [Learn more](https://www.wearedevelopers.com/videos/1222-devops-for-ai-running-llms-in-production-with-kubernetes-and-kubeflow) ## From foundation model to hosted AI solution in minutes - **Cost considerations of fine-tuning large language models** (05:26, 1min) — Fine-tuning requires massive datasets and costly specialized hardware which proves impractical for most organizational use cases. [Learn more](https://www.wearedevelopers.com/videos/1170-from-foundation-model-to-hosted-ai-solution-in-minutes) ## The LLM Evolution: From Sequence Imitation to Verifiable Reasoning - **Instruction fine-tuning and human feedback reinforcement** (12:03, 2min) — Models programmatically align toward desired communication metrics after analyzing complex behavioral outcomes guided exclusively via human rewards. [Learn more](https://www.wearedevelopers.com/videos/100343-the-llm-evolution-from-sequence-imitation-to-verifiable-reasoning) ## Coffee with Developers - Cassidy Williams - - **Designing private fine-tuning models for local codebases** (23:01, 3min) — Enterprise engineering demands customizable training models that learn from local mistakes without exposing private intellectual property. [Learn more](https://www.wearedevelopers.com/videos/912-coffee-with-developers-cassidy-williams) ## Developer Experience, Platform Engineering and AI powered Apps - **Fine-tuning and serving custom AI models** (23:14, 4min) — Automation pipelines process specific datasets to refine output behaviors and publish the resulting formats as queryable network endpoints. [Learn more](https://www.wearedevelopers.com/videos/990-developer-experience-platform-engineering-and-ai-powered-apps) ## Google Gemma and Open Source AI Models - Clement Farabet - **Customizing foundation models for specific software domains** (10:32, 1min) — Enabling developers to fine-tune base open implementations into specialized variants for targeted tasks. [Learn more](https://www.wearedevelopers.com/videos/1354-google-gemma-and-open-source-ai-models-clement-farabet) ## Multimodal Generative AI Demystified - **Fine-tuning image generation models with specific references** (26:20, 1min) — Extracting specific artistic styles without massive compute requires training textual inversion techniques on small curated datasets. [Learn more](https://www.wearedevelopers.com/videos/829-multimodal-generative-ai-demystified) ## Java Meets AI: Empowering Spring Developers to Build Intelligent Apps - **Mitigating context window limits with prompt engineering** (21:02, 1min) — Refining model inputs to produce targeted responses without relying on extensive fine-tuning. [Learn more](https://www.wearedevelopers.com/videos/1554-java-meets-ai-empowering-spring-developers-to-build-intelligent-apps) ## WeAreDevelopers LIVE – AI vs the Web & AI in Browsers - **Adopting fine-tuned task oriented AI models** (14:14, 1min) — Small fine-tuned deployment models securely solve predefined functions without demanding the overhead of sprawling general systems. [Learn more](https://www.wearedevelopers.com/videos/1743-wearedevelopers-live-ai-vs-the-web-ai-in-browsers) ## Should we build Generative AI into our existing software? - **Demystifying retrieval augmented generation and fine-tuning models** (14:24, 2min) — Explaining established architecture patterns helps stakeholders understand when to use context retrieval versus model fine-tuning. [Learn more](https://www.wearedevelopers.com/videos/1129-should-we-build-generative-ai-into-our-existing-software) ## AI That Fits Your Business, Not the Other Way Around - **Building and fine-tuning models with the NeMo framework** (07:43, 2min) — Leveraging containerized microservices operating on scalable clusters facilitates robust pre-training and custom proprietary data model fine-tuning. [Learn more](https://www.wearedevelopers.com/videos/100148-ai-that-fits-your-business-not-the-other-way-around) ## Serverless deployment of (large) NLP models - **Fine-tuning BERT models for audience sentiment analysis classification** (14:24, 3min) — Training standard language representations on domain-specific event text improves sentiment scoring accuracy for non-standard sentences. [Learn more](https://www.wearedevelopers.com/videos/158-serverless-deployment-of-large-nlp-models) ## Inside the Mind of an LLM - **Fine-tuning models to answer questions and execute tasks** (10:28, 0min) — Supervised learning on prompt and answer pairs transforms a text completion engine into a responsive assistant. [Learn more](https://www.wearedevelopers.com/videos/1617-inside-the-mind-of-an-llm) ## Your First Pitch Is to AI: How to Make Your Brand/Product Visible in the Age of Generative Search - **Fine-tuning models and publishing structured brand messaging** (00:51, 1min) — How automating message coordination across various platforms builds consistent brand context for artificial intelligence. [Learn more](https://www.wearedevelopers.com/videos/100121-your-first-pitch-is-to-ai-how-to-make-your-brand-product-visible-in-the-age-of-generative-search) ## RTX AI PC: Developing local and edge AI applications - **Cost-effective LoRA fine-tuning workflows on local inference hardware** (12:13, 1min) — Training targeted style representations locally on graphical processors drastically reduces ongoing operational costs. [Learn more](https://www.wearedevelopers.com/videos/100078-rtx-ai-pc-developing-local-and-edge-ai-applications) ## The R in RAG: Why retrieval is often the weakest link (and how to fix it) - **Preparing training datasets and executing fast model fine-tuning** (18:38, 2min) — Providing paired positive and negative examples accelerates deployment without requiring massive datasets or specialized infrastructure. [Learn more](https://www.wearedevelopers.com/videos/100005-the-r-in-rag-why-retrieval-is-often-the-weakest-link-and-how-to-fix-it)