LlamaCon 2025
who it's for
machine learning engineers, ml infrastructure engineers, platform engineers, developer-experience engineers, security engineers, data scientists
topics
Llama API hosted endpoints; fine-tuning Llama 8B; Llama 4 Maverick deployment; Llama 4 Scout single-host; speculative decoding speedups; page-KV long context; multimodal SAM object segmentation; LlamaGuard moderation models; PromptGuard prompt-injection detection; mixture-of-experts model distillation; Llama-prompt-ops and Synthetic Data Kit; dynamic quantization for deployment
| title | author | length |
|---|---|---|
| Keynote + Conversation | Mark Zuckerberg, Ali Ghodsi | 01h 07m |
Announces concrete numbers and engineering choices: 1.2 billion downloads for Llama 4 variants; a sparse mixture-of-experts design distilled from a… | ||
| Conversation | Mark Zuckerberg, Satya Nadella | 32m |
A concrete roadmap: cloud stacks must be rethought—compute, storage, network plus AI accelerators as first-class IaaS; app-server tooling (Foundry) and GitHub… | ||
| Introducing Llama API: A Seamless Way to Revolutionize Your Workflow | Dom Divakaruni | 11m |
Llama API launches a hosted path to Llama 4 Scout and Maverick with OpenAI-compatible endpoints, SDKs, and a playground for direct code export; it offers easy… | ||
| Tailoring Llama for Your Needs: Fine-Tuning & Evaluation | Michael Slater | 09m |
Two new Llama API products are introduced: a fine-tuning tool that customizes Llama 8B to specific data and an evaluations tool that measures quality… | ||
| The Llama Toolbox: A Collection of Practical Tools and Recipes | Hamid Shojanazeri | 21m |
Introduces Llama-prompt-ops for automated prompt migration and optimization (YAML-configured, runs in ~10–30 minutes) and a Synthetic Data Kit that ingests… | ||
| New Protection Tools for Llama Developers | Lauren Deason | 11m |
Meta releases concrete tools: LlamaGuard 4 is a natively multimodal moderation model that replaces text+image chaining, PromptGuard 2 ships as an 86M and a 22M… | ||
| Unpack the Latest Llama 4 Herd | RANGAPRABHU PARTHASARATHY | 12m |
Claims concrete performance and deployment details: two new models with 17 billion active parameters each—Scout is distilled to run on a single NVIDIA H100 and… | ||
| How Leidos Boosts Developer Efficiency With Llama | Beyang Liu | 02m |
Sourcegraph integrates an on‑premise Llama model into secure AWS environments to tackle large legacy codebases, replacing slow, stale open models with a… | ||
| Finetune Llama 4 Faster | Daniel Han-Chen | 03m |
Unsloth AI shows a workflow that loads and quantizes Llama 3 and 4 on Google Colab and Kaggle, integrates with Ollama, Llama.cpp, OpenWebUI and VLM, and… | ||
| nothing matches | ||