This repository provides a validated Docker-based deployment recipe for serving a DeepSeek V4 Flash variant on two DGX Spark nodes using vLLM with tensor parallelism, NVFP4 KV cache, and DSpark speculative decoding, achieving up to 1.5M context and ~50-60 tok/s throughput.
158
No data
12
4
MIT
2026-07-16
It delivers one of the first validated multi-node deployments of a 1M+ context model on consumer-grade Blackwell hardware, with comprehensive benchmarks, a patch history that resolves concurrent-serving garbling issues, and detailed tuning knobs for balancing context length and concurrency.
DeepSeek V4 Flash DSpark 1M NVFP4 KV recipe for 2x DGX Spark
Similar projects matched by category, topics, and programming language.
Harness Engineering is a methodology for improving coding agent outputs by carefully crafting the environment around them—providing curated context, tools, and executable constraints that encode an organization’s nonfunctional requirements and cumulative lessons.
A 28.9 million parameter language model runs on an $8 ESP32-S3 microcontroller entirely on-device, generating simple stories at about 9.5 tokens per second.
mimic captures traffic from any iOS or web app and automatically generates a Python client library that lets you call the app's API like a regular library.