ThinkingCap Kearuga EXL3: TandemLLM-first DGX Spark serving, independent RTX 5090 receipts and a retained TensorFold reproduction recipe.
-
Updated
Oct 19, 2026 - Python
#
reproducible-inference
Here are 3 public repositories matching this topic...
Reproducible JAX/Keras 3 + Orbax runtime and TPU v5e-8 production qualification for Microsoft Mage-Flow-Turbo, with bilingual docs, public model artifacts, and acceptance evidence.
keras kaggle image-generation text-to-image tpu jax model-conversion model-sharding bilingual-documentation orbax mage-flow tpu-v5e reproducible-inference
-
Updated
Oct 2, 2026 - Python
Portable native inference stack for Mage-Flow-Turbo using stable-diffusion.cpp/sd-cli, Q8_0 DiT GGUF, Qwen3-VL-4B Q4_K_M and a dedicated VAE, with CPU/CUDA backends, CLI, REST API, model verification and Kaggle integration.
rest-api cuda kaggle image-generation quantization text-to-image nvidia-cuda cpu-inference generative-ai safetensors gguf stable-diffusion-cpp qwen3-vl native-inference q8-0 mage-flow mage-flow-turbo qwen3-vl-4b q4-k-m reproducible-inference
-
Updated
Sep 11, 2026 - Python
Add this topic to your repo
To associate your repository with the reproducible-inference topic, visit your repo's landing page and select "manage topics."