AI News selected for Professionals and Decision Makers
Hands On Model Tooling And Research Updates

Granite 4.2 LLMs: How They're Built

17:14 · August 25, 2026 · Hugging Face Blog

Granite 4.2 LLMs: How They're Built

Summary

Granite 4.2 comprises a family of dense decoder-only transformer models released in 3B, 8B, and 30B parameter sizes. The models are trained from scratch on approximately 15 trillion tokens through a five-phase curriculum that begins with broad web-scale data and progressively shifts toward curated sources before extending the context window to 512K tokens in the final phase. All three sizes share the same architectural backbone and training pipeline, with the larger variants receiving additional post-training to support agentic behavior.

Supervised fine-tuning mixes agentic trajectories (31.6 percent) with non-agentic instruction, coding, math, and multilingual data (68.4 percent), yielding roughly 7.2 million samples after normalization to OpenAI Chat format, LLM-based quality filtering, and both local and global deduplication. A second SFT stage for the 30B model upsamples agentic coding and software-engineering trajectories while retaining a replay buffer of earlier data. The resulting checkpoints then enter a multi-stage reinforcement-learning curriculum that applies Group Relative Policy Optimization asynchronously across separate environments.

The RL pipeline sequences verifiable-reward stages (RLVR and skill boosters) for all sizes, followed by an agentic block—covering software engineering, terminal interaction, and web search—for the 8B and 30B models only. Each stage runs as an independent GRPO job that warm-starts from the preceding checkpoint, using real-environment rollouts, group-relative advantages, and truncated importance sampling to accommodate asynchronous parameter refreshes. Final alignment employs RLHF with a modest KL penalty. Native tool-calling support is preserved throughout, allowing the models to emit OpenAI-compatible function calls that integrate directly with vLLM or SGLang endpoints.

The release also includes FP8 and GGUF quantization recipes, NeMo-RL training infrastructure, and benchmarks on reasoning, software-engineering, and agentic tasks. All variants are distributed under the Apache 2.0 license and incorporate Dutch-language coverage within the multilingual portion of the SFT mixture.

Why it matters

Provides production-grade details on training pipelines, RL methods, memory/latency optimizations, and deployment that ML engineers can directly apply or replicate in Dutch/EU settings.

More in this beat
Wire It, Run It, Deploy It: AI Workflows in Gradio

02:00 · August 25, 2026

Wire It, Run It, Deploy It: AI Workflows in Gradio

It provides ML Engineers with a highly actionable, hands-on tool for rapid prototyping and deploying AI workflows. The automatic REST API generation and seamless GPU integration streamline the transition from model testing to accessible endpoints, which is highly valuable for agile AI teams and SMEs.

Relevance 65 · Audience 85

Measuring benchmark optimization in speech recognition

02:00 · August 21, 2026

Measuring benchmark optimization in speech recognition

Directly addresses evaluation metrics, benchmark reliability, and real-world generalization for ML engineers selecting ASR models; includes EU parliamentary data and actionable advice on avoiding over-optimistic scores.

Relevance 60 · Audience 75

Up to 3.2x Faster Inference with LFM2.5-DSpark

18:52 · August 20, 2026

Up to 3.2x Faster Inference with LFM2.5-DSpark

Directly addresses production inference challenges like memory-bound decode latency and GPU/edge deployment for ML Engineers, with quantitative benchmarks and open implementations applicable in Dutch AI workflows.

Relevance 85 · Audience 90

How Much Memory Does Your Agent Actually Need?

20:09 · August 18, 2026

How Much Memory Does Your Agent Actually Need?

This article provides highly actionable, production-focused insights for ML Engineers building AI agents. It addresses critical MLOps challenges like balancing inference cost with model accuracy through prompt caching and dynamic context retrieval, which is highly applicable for Dutch tech teams optimizing LLM deployments.

Relevance 85 · Audience 95

Multi-Vector (Late Interaction) Embedding Models with Sentence Transformers

02:00 · August 18, 2026

Multi-Vector (Late Interaction) Embedding Models with Sentence Transformers

Directly actionable for ML Engineers building production retrieval systems: addresses latency, VRAM/index tradeoffs, distributed setups via vector DBs, quantitative benchmarks, and domain-specific edge cases like long documents or visual pages. Fully applicable to Dutch AI teams via open-source tooling.

Relevance 85 · Audience 90

Same Cluster, 33 Points More Utilization: What Changed Was the Order

21:46 · August 17, 2026

Same Cluster, 33 Points More Utilization: What Changed Was the Order

Directly addresses production GPU orchestration challenges (contention, reservations, contiguous blocks, churn) with quantitative benchmarks and implementation details relevant to ML engineers running mixed training/inference workloads on shared hardware.

Relevance 78 · Audience 85

Thinking of ACE? We Can Do It with Fewer Tokens

15:37 · August 11, 2026

Thinking of ACE? We Can Do It with Fewer Tokens

This article provides actionable insights for ML Engineers building LLM agents, offering a concrete method (ALTK-Evolve) to reduce inference costs and token usage without sacrificing accuracy. It directly addresses production challenges like context overload and compute efficiency, which are critical for Dutch enterprises scaling AI solutions.

Relevance 85 · Audience 95