Topic
Gemma 4
3 episodes
-
1Aggregate results comparing DFM Mimir 1B against the HRM Text 1B, Qwen 3.5 2B and Gemma 4 E2B, displaying highly competitive performance across 20 benchmarks.
Pippa and Tyler examine DFM Mimir v1, a one-billion-parameter HRM trained from scratch with permissible post-training data, and ask whether its strong benchmark results make it genuinely useful for low-resource language work.
-
Introducing Apex: A Fast, Specialized Model for React Native
Cody and Justy dig into Callstack's Apex, a specialized React Native coding model built on Gemma 4. Cody pushes on the self-reported benchmarks, the 'private beta with our own engineers' problem, and whether 'specialized' is real or just branding. Justy defends the economic logic—GitHub Copilot's billing shift proves general models are expensive—and argues that React Native's genuine cross-platform constraints make it a real candidate for specialization. They find middle ground on where Apex might actually earn its place versus where the claims outpace the evidence.
-
How to Implement Tool Calling with Gemma 4 and Python MachineLearningMastery
Episode 287 of Exploring Next dives into the world of tool calling with Gemma 4 and Python, exploring how to build a local, privacy-first tool-calling agent.