Tag
nvfp4
Also taggeddgx-sparkvllmdflashmixture-of-expertsspeculative-decodingabliterationuncensoredblackwell
Posts and projects tagged nvfp4
- The AEON Forge catalog: every project and what to read first
A guided tour of the AEON Forge catalog: NVFP4 models for the DGX Spark, MLX builds for Apple Silicon, the voice stack, the media tools, the Orb and the Bench, with the article to read first in each group.
The Aeon Forge11 min read - Qwen 3.8 27B AEON Ultimate: the current flagship and how to serve it
Qwen3.8-27B AEON Ultimate Uncensored is the model I now serve on one Spark, two Sparks and the RTX box: what coherent abliteration removes and keeps, the NVFP4-MIXED weight lattice, the three seat recipes with their commands, and the rules that keep them running.
The Aeon Forge14 min read - Serving Qwen 3.6 35B-A3B heretic at NVFP4 with DFlash on a DGX Spark
How I serve an uncensored Qwen 3.6 35B-A3B at NVFP4 with DFlash speculative decoding on one DGX Spark: the seven container fixes, the mixture-of-experts angle, the drafter, the measured numbers and the exact command that works today.
The Aeon Forge11 min read - Ornith 1.0 35B AEON Ultimate keeps the coding and drops the refusals
I abliterated DeepReinforce's Ornith 1.0 35B, an agentic coding MoE, and kept the coder: 0 of 80 refusals, pass@1 identical to the base, a three-way BF16, NVFP4 and DFlash benchmark on a DGX Spark, and the quick start to serve it.
The Aeon Forge10 min read - Nemotron 3 Nano Omni AEON Ultimate, served uncensored on a DGX Spark
An abliterated Nemotron 3 Nano Omni that reasons over text, images and audio on one DGX Spark. How the twelve-dimension refusal removal was found, what NVFP4 costs against BF16, what the source-built vLLM image patches, and how to deploy it.
The Aeon Forge9 min read - modelopt-fast-moe cuts AWQ calibration time for MoE models at NVFP4
modelopt-fast-moe replaces the one-sample-at-a-time calibration loop that leaves a GPU mostly idle while nvidia-modelopt quantizes a Mixture-of-Experts model to NVFP4. What broke on a 26B MoE, what the batched loop fixes, and what it still leaves to you.
The Aeon Forge6 min read - Gemma 4 31B DECKARD HERETIC at NVFP4 with DFlash on a DGX Spark
A dense, uncensored Gemma 4 31B with thinking, quantized to NVFP4 and served through a DFlash drafter on one DGX Spark. Single-stream decode went from 11 to 39 tokens per second, 32 concurrent requests reach 427, and reasoning and tool calling stay intact.
The Aeon Forge11 min read - Gemma 4 26B-A4B at NVFP4: a mixture-of-experts recipe for a DGX Spark
How I serve Gemma 4 26B-A4B, a mixture-of-experts model at NVFP4, on one DGX Spark with a DFlash drafter: the production recipe, the bench recipe behind 816 tok/s at c64, and the drafter defects that made my earlier numbers wrong.
The Aeon Forge10 min read - ComfyUI for the DGX Spark: the AEON media production base
A Docker build of ComfyUI for the DGX Spark with CUDA 13, SageAttention v3 for sm_121a and NVFP4 weights, pulling Flux 2, LTX 2.3 and ACE-Step on first start. What broke, what the flags mean, how local and remote modes work, and why three AEON media repos sit on it.
The Aeon Forge10 min read - AEON vLLM Ultimate RTX: vLLM for the RTX 5090 and Blackwell sm_120
A slim vLLM image for the RTX 5090 and RTX PRO 6000 Blackwell, why it must never be crossed with the DGX Spark image, what changes on a 32 GB card, four serving recipes for a 27B NVFP4 model, and the WSL2 fault that passed every test.
The Aeon Forge9 min read - AEON vLLM Ultimate: one serving image for the DGX Spark
AEON vLLM Ultimate is vLLM built from source for the DGX Spark's GB10 and shipped as one 18.3 GB image that serves the whole AEON fleet with speculative decoding and NVFP4 weights. Here is why it exists, what broke on the way, and how to pull and run it.
The Aeon Forge14 min read