NVIDIA A100

Headlines · 4

1. 6d
   Glance Forcing cuts video diffusion distillation to one trajectory

   An arXiv paper (arXiv:2610.20220, submitted 14 Oct 2026) proposes Glance Forcing, a lightweight distillation framework that converts bidirectional diffusion video models into causal few-step autoregressive students. The authors say existing pipelines rely on offline Probability Flow ODE trajectories costing tens of thousands of GPU-hours and terabytes of storage, and that reducing supervision to a single sample causes quality collapse traced to differing curvature in the high-noise and low-noise regimes. Glance Forcing instead decouples the denoising trajectory into slow and fast experts with stage-specific objectives while keeping the pretrained teacher frozen; the paper reports that with only one PF-ODE trajectory it raises VBench Total from 84.04 to 84.22, cuts trajectory-collection cost from thousands of GPU-hours to about one A100 GPU-hour, and shows larger gains in dynamic degree, VisionReward and instruction-following, with human raters consistently preferring it over strong baselines.

   Hugging Face · Papers · 🔥 0
2. 6d
   Modal tests mixed-precision, mixed-accelerator RL samplers on Qwen3.6-35B-A3B

   Modal published an engineering post asking whether large-model reinforcement learning can use samplers that run different precisions and accelerators than the trainer, each with its own kernels. Its mixed samplers span six accelerator types — B200 and B300 serving NVFP4 weights, H100 and H200 serving FP8 and BF16, and A100 and RTX PRO 6000 serving BF16 — all with FP8 KV caches, feeding a single BF16 trainer on B200. Modal says off-policy correction plus autoscaling serverless GPUs let it train Qwen3.6-35B-A3B fully asynchronously on agentic coding tasks with up to 256K-token contexts, reporting that the mixed samplers' mismatch averaged 4.7× that of homogeneous samplers.

   Modal Blog · 🔥 0
3. Oct 11
   Microsoft open-sources TRELLIS.2, a 4B image-to-3D model with PBR output

   Microsoft has published TRELLIS.2 on GitHub, a 4-billion-parameter image-to-3D generative model built on a "field-free" sparse voxel representation called O-Voxel, which the repo says handles open surfaces, non-manifold geometry and internal enclosed structures and predicts PBR surface attributes such as base color, roughness, metallic and opacity. The project reports generation times on an NVIDIA H100 of about 3 seconds at 512³ (2s shape + 1s material), roughly 17 seconds at 1024³ and about 60 seconds at 1536³, plus under 10 seconds on a single CPU to convert a textured mesh to O-Voxel and under 100ms on CUDA for the reverse. Code, data tooling and training scripts are provided and the TRELLIS.2-4B weights are on Hugging Face, but the code is Linux-only and needs an NVIDIA GPU with at least 24GB of memory (verified on A100 and H100, CUDA 12.4).

   GitHub Trending · Python · 🔥 0
4. Oct 7
   SparseDecoding: decoding-aware pruning for faster LLM inference

   The paper introduces SparseDecoding, a training-free network pruning framework aimed at the decoding stage of LLM inference. It builds calibration matrices from layer-wise activations collected during dense-model autoregressive generation (excluding prefill) so the pruning objective matches decoding activations, and pairs this with an N:M sparse matrix-vector kernel using bitmask indexing and fixed-step traversal. The authors report that on Llama-3.1-8B, Llama-3.3-70B and Qwen3-14B/32B it outperforms fixed-text calibration on long-form generation benchmarks, with up to 1.48x end-to-end wall-clock decoding speedup on A100 GPUs.

   Hugging Face · Papers · 🔥 0

Experience and discussion from the community

Share my NVIDIA A100 experienceAsk about NVIDIA A100