Tag
Explore by tags
Generates streaming, low‑latency neural speech for real‑time dialogue by autoregressively producing audio frames as text arrives; joint text–speech training preserves natural prosody. Optimized for vLLM streaming (~50 ms first chunk), supports short‑clip voice cloning and four languages.
Provides a pre-quantized NVFP4 checkpoint of GLM-5.2 for long-context reasoning and coding; reduces model footprint so GLM-5.2 can run on multi‑GPU Blackwell nodes and is ready for inference with SGLang and vLLM.
NVFP4-quantized variant of Qwen3.6-27B that reduces parameter bits from 16 to 4, cutting disk and GPU memory requirements by ~2.5× while keeping comparable benchmark accuracy; ready for vLLM-based inference on NVIDIA hardware and supports long, multimodal contexts.
Deployment-optimized hybrid MoE LLM (75B total / 9.3B active) produced via Iterative Puzzle compression and Multi-Token Prediction to double server throughput and raise single-GPU concurrency; designed for multilingual reasoning, long-context generation, and high-volume agentic/chat deployments.
Generates 2048-d multilingual text embeddings for retrieval and semantic search, suited for RAG and dense retrieval. Pruned and distilled from the Ministral-3 family into a ~1.14B BF16 model, supports long contexts (up to 32,768 tokens) and optimized for NVIDIA GPU inference.
Adds vision to GLM-5.2 by attaching a MoonViT encoder and a trained 49.5M-parameter PatchMerger projector to enable image→text multimodal reasoning; text and vision backbones are frozen, uses NVFP4 quantized weights and targets Blackwell B200 GPUs.
Generates and reasons about multimodal physical-world content—text, images, video and action trajectories—conditioned on text, images, video and robot/vehicle action inputs. An edge-sized (4B) Mixture‑of‑Transformers omni-model optimized for single‑GPU inference and Physical AI tasks (image→video, action prediction, robot policy).
4-bit NVFP4 (W4A4) quantized pack of Upstage Solar Open2 250B for vLLM serving on NVIDIA Blackwell GPUs, preserving MoE routing and near-BF16 quality while cutting model size from 500.6 GB to 153.3 GB.
A PyTorch-native training framework for agentic reinforcement learning research that keeps researcher-facing code compact and editable. Uses an asynchronous loop to train multimodal and mixture-of-experts policies while never training on tokens the agent didn't generate; matches Megatron-style stacks under a comparable protocol and ships recipes and containers on GitHub.
Consolidated dataset of detection, visual grounding and pointing annotations with indexed WebDataset image shards and Megatron‑Energon training metadata. Covers diverse visual domains (COCO, RefCOCO, driving, GUI, documents) and uses a normalized spatial grid for cross‑domain vision–language grounding training.
An end-to-end 11B full-duplex speech model for real-time conversational AI that jointly performs streaming speech understanding and generation, enabling ~450 ms turn-taking, barge‑in and live tool calling in a single unified architecture; research use only.
Directly maps visual observations and language instructions to continuous robot actions, replacing LLM-centric V→L→A pipelines. Uses separate visual and language encoders with lightweight bidirectional interaction and a compact decoder to cut inference cost and VRAM, achieving ~31 ms latency and <1 GB VRAM on an RTX 4090; suited for real-time robotic manipulation under tight compute budgets.