#speculative-decoding (17 Repositories)
Ranked open-source repositories tagged with #speculative-decoding, scored by pull request acceptance likelihood and maintainer engagement velocity.
55.3%
11.0h
17 repositories tagged #speculative-decoding
Avarok-Cybersecurity/atlas
Pure Rust Inference Engine
avifenesh/bw24
From-scratch Rust+CUDA inference engine, bit-exact by construction — NVFP4, MoE, MTP speculative decoding, tuned against measured limits of one RTX 5090 Laptop (sm_120a).
StayLameBro/backburner
Your iPhone helps your Mac run a 27B model: faster prompt reading and more context over a USB-C cable
Tencent/AngelSlim
Model compression toolkit engineered for enhanced usability, comprehensiveness, and efficiency.
SergiioB/intel-arc-pro-b70-inference-cookbook
Open recipes, engine patches, and benchmark harnesses for LLM inference on Intel Arc Pro B60/B70 (Battlemage, Xe2). MoE 35B at 160 t/s decode / 7.5K t/s prefill single-stream, 27B at 50~ t/s decode / 1.7K t/s prefill single stream. vLLM XPU MTP unlocked. Muse Glimmer recipe added!!
syv-ai/qwen38-27b-rtx3090
Qwen3.8-27B on a single RTX 3090 with vLLM: ~1,000 tok/s at 64 concurrent (int8 tensor-core GEMMs, fp16 DeltaNet state), ~114 tok/s single-user at default sampling / ~124 greedy (MTP drafts, own-output draft vocab, calibrated int4 lm_head, split-KV verify attention), 150k-262k context; patches, requant scripts, benchmarks
dphnAI/sonar
Large-scale LLM inference engine
incoai/splash
A local inference engine for Apple silicon, built around the model.
swellweb/reame
CPU-first LLM inference server on llama.cpp. Runs useful models on free-tier ARM boxes; rewriting the input made it ~6x faster and more accurate than tuning the engine. MIT, benchmarks and failures included.
julianmb/q38rocm
Qwen 3.8 27B ROCmFP4 on AMD Strix Halo (Ryzen AI Max+ 395). Up to 36 tok/s via MTP Speculation, TurboQuant & Mesa RADV Wave64.
syv-ai/HyperQwen
Serve large Qwen models fast on the GPUs you actually own. Qwen3.8-27B on a single 24 GB card with vLLM: 127 tok/s single-user (381 when the answer quotes the prompt), ~1,035 tok/s at 64 concurrent, 150k-262k context. vLLM patches, requant pipeline, benchmarks.
Nehanth/pooled
Peer-to-peer LLM inference in the browser: pool your devices to run big open models, for chat and coding agents.
avifenesh/memra
Rust + CUDA inference engine for NVIDIA RTX PRO 6000 Blackwell and RTX 5090. Serves safetensors and GGUF over an OpenAI-compatible API, with per-device tuned defaults and speculative decode gated byte-identical to plain decode. Hosted instance: inference.tiyuvta.ai
hasso5703/dgx-spark-qwen38
Fastest measured Qwen3.8-27B config for DGX Spark (GB10): SGLang + NVFP4 + DFlash2, 72 tok/s greedy median, 1M context, deterministic boots. Flash-Next 176B on one box. One command, everything pinned, and a local cockpit to drive it.
Nehanth/swarmllm
Every device brings a slice. Together they run the whole model. Peer-to-peer LLM inference across browser tabs: a from-scratch WebGPU engine and a WebRTC runtime that split a 27B model over the devices in a room.
Nathanw1014/strix-halo-llamacpp
Performance-tuned llama.cpp for AMD Strix Halo (gfx1151): FA + MoE-prefill fixes with a bundled current Mesa driver. Vulkan and HIP; portable dir, Docker, and distrobox.
AEON-7/Qwen3.6-27B-AEON-Ultimate-Uncensored-DFlash
Fully uncensored, capability-enhanced abliteration of Qwen3.6-27B. NVFP4 + z-lab DFlash speculative decoding (n=12) on the unified ghcr.io/aeon-7/aeon-vllm-ultimate:latest container, tuned for long-context draft acceptance on DGX Spark. 6 HF variants (BF16/NVFP4/MTP/MTP-XS), docker-compose, and QuickStart.