PyTorch

Latest developments

End-to-End Multimodal Data Augmentation and Adversarial Robustness Benchmark with AugLy for Images, Text, Audio, and PyTorch MarkTechPost · 4 hours ago · 46 NVIDIA Details BioNeMo Inference Runtime (BioIR): 2.90x Higher Boltz-2 Folding Throughput and 58.5K Residues per GPU-Hour on 8xH100 MarkTechPost · 2 weeks ago · 14 Fine-Tuning Tool-Calling LLMs: A Complete Guide Using XYZ-Aquila-SFT and Qwen3 MarkTechPost · 1 month ago · 48 AllenAI Open Instruct Tulu 3 Post-Training with SFT, DPO, RLVR, GRPO, and Verifier-Based Evaluation MarkTechPost · 1 month ago · 44 Xiaomi’s MiLM Plus Releases PROVE: Perception-Aligned Object Removal Metrics RC-S and RC-T With a Real-World Video Benchmark MarkTechPost · 1 month ago · 1

Cursor Open-Sources Mixture-of-Kittens (MoK): A Deterministic MoE Training Megakernel for GB300 NVL72 Racks MarkTechPost · 1 month ago · 46

Q3 2026

◆ Cursor Research Open-Sources Mixture-of-Kittens Deterministic MoE Training Megakernel Open source release

• 26 Sep End-to-End Multimodal Data Augmentation and Adversarial Robustness Benchmark with AugLy for Images, Text, Audio, and PyTorch
• 10 Sep NVIDIA Details BioNeMo Inference Runtime (BioIR): 2.90x Higher Boltz-2 Folding Throughput and 58.5K Residues per GPU-Hour on 8xH100
• 15 Aug Fine-Tuning Tool-Calling LLMs: A Complete Guide Using XYZ-Aquila-SFT and Qwen3
• 12 Aug AllenAI Open Instruct Tulu 3 Post-Training with SFT, DPO, RLVR, GRPO, and Verifier-Based Evaluation
• 12 Aug Xiaomi’s MiLM Plus Releases PROVE: Perception-Aligned Object Removal Metrics RC-S and RC-T With a Real-World Video Benchmark
• 4 Aug Cursor Open-Sources Mixture-of-Kittens (MoK): A Deterministic MoE Training Megakernel for GB300 NVL72 Racks

Relationships

Products & technology

• Integrated with Qwen3-0.6B · 1 source
• Mixture-of-Kittens integrated with this tool · 1 source
• PROVE integrated with this tool · 1 source
• BioNeMo Inference Runtime (BioIR) integrated with this tool · 1 source
• AugLy integrated with this tool · 1 source