Featured
AI infrastructure
New 2026
A high-throughput and memory-efficient inference and serving engine for LLMs
New 2026
A context engine for DeepSeek Harness — reversible, token-budgeted compression of the live context window.
New 2026
Qwen3.8-27B full-NVFP4 deployment on 2x NVIDIA DGX Spark with SGLang, RoCE, NGRAM/MTP, and benchmarks
New 2026
Local-first, open-source experiment tracking for machine learning and agents
New 2026
Connect your ChatGPT, Claude, GitHub Copilot, Grok, and OpenCode subscriptions and use them securely from local apps, APIs, SDKs, and a see them via dashboard.
New 2026
Qwen3.8-27B at 34-38 tok/s on DGX Spark (GB10) — one-command SGLang + NVFP4 + DSpark setup, systemd, Claude Code ready
New 2026
Qwen3.8-27B on a single RTX 3090: 416 tok/s batched or 25 ms/token single-user, 150k context, vLLM
New 2026
Official Cursor SDK to Anthropic-compatible HTTP gateway with native MCP tool continuation.
New 2026
Self-hosted GPU cluster orchestration: deploy and serve AI models on your own hardware
New 2026
Serving Qwen3.8-27B (Unsloth UD-Q4_K_XL GGUF) at 23 tok/s over a 104,192-token context on a single NVIDIA L4 24GB, llama.cpp + MTP speculative decoding. One-command GCP provisioning, tuning sweep, quality verification, and a binary-searched context ceiling.
New 2026
use glm/minimax/openai/claude api in your deepseek harness
New 2026
Logical KV cache memory profiler, zombie leak detector, and fragmentation inspector for PagedAttention (vLLM, SGLang)
New 2026
Reproducible Docker deployment for Qwen3.8-27B NVFP4 on NVIDIA DGX Spark (GB10, SM121) via OpenAI-compatible vLLM. Known-good baseline + guarded redeploy/rollback, diagnostics, and benchmark. Unofficial community project.
New 2026
🔥 FTRAIN V1.0 — High-performance AI training framework designed for maximum speed, scalability, and memory efficiency. Streamline deep learning workflows, model pre-training, and fine-tuning with low-overhead distributed computing and optimized data pipelines. Built for high-throughput ML workloads.
New 2026
Safety protocol framework for LLM agents. Binding, scope, budget, approval, monitoring, audit, kill switch. Reference architecture with on-chain binding, on-chain audit, and insurance interface.
New 2026
Gemma 4 31B FP8 MTP5 vLLM profile for two Intel Arc Pro B70 GPUs
New 2026
Curated community plugin directory and live marketplace for DeepSeek Harness.
New 2026
Native Muse Glimmer 30B runtime for two Intel Arc Pro B70 GPUs
Other topics
Browse other topics on Trendshift
A
N