Eldar Kurtić (@_EldarKurtic) / X

Our DASLab group's models are breaking Hugging Face Hub's download counters! Great work @mhelcig

Run a 763B-parameter model locally. DeepSeek-V4.1-Flash-1017 on a 12GB GPU. Frontier quality with GSQ-RCO quants, directly supported in the Strata inference engine. Further info below. Much more to come. Follow for updates.

Tech influencers are the best thing that ever happened to this platform.

You’re unable to view this Post because this account owner limits who can view their Posts. Learn more

NVIDIA just released Nemotron 3.5 Super VL in BF16 and NVFP4, so we are complementing their release with FP8 checkpoint optimized for Hopper deployments and a DSpark speculator for even faster inference than the built-in MTP. You can even combine them for compounded gains!

Red Hat AI

@RedHat_AI

Oct 15

Two optimized checkpoints for NVIDIA Nemotron 3.5 Super VL, a 120B multimodal MoE for long-running agents. Quantization and speculative decoding stack for close to 2x on Hopper, accuracy intact. FP8 (LLM Compressor): huggingface.co/RedHatAI/NVIDI… DSpark (Speculators):

Join us for a chat about speculative decoding! We'll share some practical tips and tricks for both training and deployment in production.

Hugging Face

@huggingface

Oct 14

Learn about how to get multipliers in local LLM speed-ups using speculative decoding in this livestream with @mervenoyann, @MaartenGr and @_EldarKurtic youtube.com/watch?v=H4azmW…

More state-of-the-art MXFP4 ckpts ready for inference on AMD GPUs and @vllm_project available at: huggingface.co/RedHatAI/Qwen3…

2

Eldar Kurtić

Eldar Kurtić’s posts