Autonomous self-improving 4x DGX Spark (GB10) MoA stack + LoRA loop (DSV4F router, Qwen3.6/Omni/TwoTower/Gemma). Hermes MoA routing, ~90% local, self-LoRA. Includes a SINGLE-SPARK recipe (all-Gemma-4: 31B brain + DiffusionGemma + 12B) to build the same on 1 GB10.
☆24Aug 17, 2026Updated 2 months ago
Alternatives and similar repositories for Keys-Setup-Autonomous-Self-Improving-Local-Inference-Stack
Users that are interested in Keys-Setup-Autonomous-Self-Improving-Local-Inference-Stack are comparing it to the libraries listed below. We may earn a commission when you buy through links labeled 'Ad' on this page.
Sorting:
- MiMo-V2.5 NVFP4 4-bit weights + NVFP4 4-bit KV cache + DFlash speculative decoding on 2× NVIDIA DGX Spark — 1M context, 3.4M-token KV poo…☆18Jul 13, 2026Updated 3 months ago
- GLM-5.2 Quantrio INT4/INT8 Mixed Abliterated — SPEED=1 C1≈30 tok/s @ 128k on 4x DGX Spark. Step-by-step recipe, image bake, results.☆13Jul 18, 2026Updated 3 months ago
- Universal AI computer-control KVM (REST API + MCP) that drives any machine as plain USB hardware — nothing installed on the target. Built…☆20Updated this week
- GLM-5.2 QuantTrio TP=4+DCP2 on 4x NVIDIA DGX Spark (GB10) — tuned and measured under a real agent workload. Adaptive MTP, the indexer law…☆20Aug 24, 2026Updated 2 months ago
- Run DeepSeek-V4-Flash at full 1M context AND two MiniMax H3 video instances on the same two DGX Sparks. Benchmarked C1-C6.☆36Aug 7, 2026Updated 3 months ago
- Bare Metal GPUs on DigitalOcean Gradient AI • AdPurpose-built for serious AI teams training foundational models, running large-scale inference, and pushing the boundaries of what's possible.
- MiniMax-H3 on single DGX Spark: keys Sol-Engine ports + Heretic TE + Spectrum/Sage speed + 0.5MP upscaler finish. Formal bench included.☆51Aug 23, 2026Updated 2 months ago
- Run on TWO-DGX-Spark - vLLm-0.24.0 dual cache optimized DSV4F+DSpark+NVFP4 KV (Concurrency 12 with 1.5M context/3M KV token Pool) >0.58-0…☆21Aug 17, 2026Updated 2 months ago
- DeepSeek-V4-Flash-0731 tuned on 4x DGX Spark (GB10) at TP=4 — 123.13 tok/s decode. Every variable measured one at a time, with the values…☆13Aug 11, 2026Updated 3 months ago
- MiniMax-M3 (428B, no pruning) at 36 tok/s on 2× NVIDIA DGX Spark — W4A16 GPTQ + NVFP4 KV + EAGLE-3 speculative decoding on vLLM. Three se…☆45Jul 13, 2026Updated 3 months ago
- Working recipe to serve DeepSeek-V4-Flash across two NVIDIA DGX Spark (GB10) nodes with vLLM (TP=2, FP8 KV, MTP) over a RoCE/RDMA link — …☆84Jul 13, 2026Updated 3 months ago
- Eight-system DGX Spark engineering: fabric repair, NCCL tuning, rack recabling, custom 3D-printed PSU supports, and thermal validation.☆11Aug 3, 2026Updated 3 months ago
- AEON vLLM Ultimate — vLLM 0.27.1 built from source for DGX Spark / Blackwell (sm_121a/GB10). DSpark quantized Markov heads, DFlash SWA on…☆149Updated this week
- Concurrency Patch to enable on DSpark - DeepSeek V4 Flash/Pro C=16 can be modified for more☆12Jun 29, 2026Updated 4 months ago
- GLM-5.3-Flash (Z.ai) NVFP4 (LibertAI) + NVFP4 KV, 1M context on 2x DGX Spark GB10 — vLLM TP2 recipe, walls ledger, benchmarks, needle to …☆29Aug 31, 2026Updated 2 months ago
- Deploy to Railway using AI coding agents - Free Credits Offer • AdUse Claude Code, Codex, OpenCode, and more. Autonomous software development now has the infrastructure to match with Railway.
- Recipe: GLM-5.2 (unpruned QuantTrio Int4-Int8Mix) at 200K ctx with MTP spec decode on a 4x NVIDIA DGX Spark (GB10) cluster☆77Jul 25, 2026Updated 3 months ago
- English guide + one-command patcher for the vLLM spin-wait bug that overheats NVIDIA DGX Spark / GB10. Credit: nacyot. CPU cores spin at …☆24Aug 17, 2026Updated 2 months ago
- We gave AI models telepathy.☆27Sep 25, 2026Updated last month
- Hermes Agent benchmark v0.1 — evaluate local models on the actual tool-calling patterns Hermes users hit. Reproducible, trace-capturing, …☆17Jun 24, 2026Updated 4 months ago
- keys-DGX-Sparkticus Utimate Power Pack Unleashed (Dual DGX-Sparks Needed)☆44Aug 23, 2026Updated 2 months ago
- Headless / multi-user serving-mode for the DGX Spark: free unified memory for vLLM by paring back the desktop. Companion to qwen3.5-122B-…☆21Jun 29, 2026Updated 4 months ago
- DeepSeek V4 Flash @ 1M token context on 2x NVIDIA DGX Spark — production-tested recipe (45 tok/s decode, real 800K prompts served)☆37Jul 13, 2026Updated 3 months ago
- Recover usable memory on DGX Spark with a 64 KiB kernel: an opt-in NVIDIA UVM page-table packing patch, build tools, and measured validat…☆39Oct 16, 2026Updated 3 weeks ago
- Honeycomb Lab — hex map + OpenAI gateway control plane for a home AI fleet☆17Oct 26, 2026Updated 2 weeks ago
- Open source password manager - Proton Pass • AdSecurely store, share, and autofill your credentials with Proton Pass, the end-to-end encrypted password manager trusted by millions.
- MiMo-V2.5 Omni TP=2 on 2x DGX Spark · 1M context · NVFP4 4-bit KV (~1.97M-token KV pool @ 1M, ~30 tok/s) · 69-eval: thinking-OFF 97.8 bea…☆39Jul 13, 2026Updated 3 months ago
- Three-DGX-Spark TP3 serving recipe for GLM-5.3 Flash☆21Oct 5, 2026Updated last month
- keys-latest: GLM-5.2 QuantTrio INT4/INT8 ablit + DFlash · 4x DGX Spark · k=12 · 128k · seqs=4 · KV 10GiB☆14Aug 17, 2026Updated 2 months ago
- A curated list of awesome local AI projects and repos☆52Updated this week
- macOS Hermes skill for sharing one canonical writable profile across Buzz and ACP surfaces☆54Jul 30, 2026Updated 3 months ago
- keys- 1M context: Inkling-Small-NVFP4 + DSpark + NVFP4 KV on SGLang, sm_121a-optimized on Two DGX Sparks — one-shot agent-executable rec…☆10Aug 17, 2026Updated 2 months ago
- CloudlessOS☆14Aug 20, 2026Updated 2 months ago
- Companion script for Hermes Agent ComfyUI skill☆17May 13, 2026Updated 5 months ago
- Give the DeepSeek Harness (dsh) eyes with any text model + any vision model: a vision proxy for chat image attachments and an analyze_ima…☆12Sep 28, 2026Updated last month
- Deploy to Railway using AI coding agents - Free Credits Offer • AdUse Claude Code, Codex, OpenCode, and more. Autonomous software development now has the infrastructure to match with Railway.
- Switchless RoCE collectives, NCCL profiles and fabric tooling for DGX Spark inference recipes☆43Updated this week
- ☆16Sep 25, 2026Updated last month
- GLM-5.2-Int4-Int8 on 8x GB10 cluster☆39Jul 30, 2026Updated 3 months ago
- Agentic test batteries for local models via Hermes Agent - scripted tools + real Hermes CLI sessions☆15Aug 14, 2026Updated 2 months ago
- Plain-language web remote control for a fleet of ComfyUI instances (Qwen-Image-2.1 stills, MiniMax-H3 video). Stdlib Python, no dependenc…☆47Oct 9, 2026Updated last month
- Wrap commands that might cause out-of-memory errors, and stop them before they freeze your machine☆15Sep 22, 2026Updated last month
- Mixed-capability LLM benchmark for DGX Spark — 57 scenarios, 10 domains, partial-credit grading, trial statistics☆171Oct 13, 2026Updated 3 weeks ago
Are these results useful?