Commits · e3bb5c46c96ba6863a62d38198adcc2fd48c275a · nexlab / coderai · GitLab

Oct 28, 2026
- choose the dtype default from a device that has been detected · 9646dd69
  Stefy Lanza (nextime / spora ) authored Oct 28, 2026
  9646dd69
- refuse a quantized model the card cannot hold, instead of proving it slowly · d94b144e
  Stefy Lanza (nextime / spora ) authored Oct 28, 2026
  d94b144e
Oct 27, 2026
- normalise a language before transcribing, stop inflating weights that are... · 3f9b2680
  Stefy Lanza (nextime / spora ) authored Oct 27, 2026
  3f9b2680
Oct 17, 2026
- nothing on the event loop may block, and the loop now says when it did (v0.4.17) · 174a3753
  Stefy Lanza (nextime / spora ) authored Oct 17, 2026
  174a3753
Sep 17, 2026
- llama pod on the light core: the GGUF backend asks NVML about the card, not torch · 4866b69d
  Stefy Lanza (nextime / spora ) authored Sep 17, 2026
  4866b69d
Sep 14, 2026
- text: LoRA and QLoRA on text models, locally and on RunPod · 34d08890
  Stefy Lanza (nextime / spora ) authored Sep 14, 2026
  34d08890
Jun 24, 2026
- nvidia HF loader: parity with hf_loading (precision, flash key, buffers, RAM cap) · 8c1ea33d
  Stefy Lanza (nextime / spora ) authored Jun 24, 2026
  8c1ea33d
- hf: resolve relative offload_dir to the configured dir; forward HF_TOKEN · fe7da690
  Stefy Lanza (nextime / spora ) authored Jun 24, 2026
  fe7da690
Jun 22, 2026
- engine: route hot-path logs through os.write (fast_print) instead of print() · e2db1b5c
  Stefy Lanza (nextime / spora ) authored Jun 22, 2026
  e2db1b5c
Jun 18, 2026
- feat: smart context caching, VRAM offload fix, GPTQ/AWQ quant backend · 990f9471
  Stefy Lanza (nextime / spora ) authored Jun 18, 2026
  990f9471
- front/engine split, ds4 + media tooling, gemma-4 native tools; ignore runtime artifacts · b297b25f
  Stefy Lanza (nextime / spora ) authored Jun 18, 2026
  b297b25f
Jun 10, 2026
- Backends, API, and tooling updates; gitignore township_output · f21c6185
  Stefy Lanza (nextime / spora ) authored Jun 10, 2026
  f21c6185
May 07, 2026
- Add prompt caching and prompt aggregation · 07fb3251
  Stefy Lanza (nextime / spora ) authored May 07, 2026
  
  07fb3251
May 05, 2026
- Multimodal capabilities · 1a723602
  Stefy Lanza (nextime / spora ) authored May 05, 2026
  
  1a723602
Mar 20, 2026
- Add --offload-strategy none to disable CPU offloading and VRAM auto-detection · beded066
  Your Name authored Mar 20, 2026
  beded066
- Add --no-ram option to maximize VRAM usage · b782a092
  Your Name authored Mar 20, 2026
  b782a092
Mar 18, 2026
- Fix: Add repeat_penalty, presence_penalty, frequency_penalty params to NvidiaBackend · 4249e178
  Your Name authored Mar 18, 2026
  4249e178
- feat: Pipeline fixes, regex optimization, GBNF grammar support, and prompt distillation · 5341ee6a
  Your Name authored Mar 18, 2026
  5341ee6a
Mar 17, 2026
- Add debug output for flash-attention and force-reasoning mode · b7d84534
  Your Name authored Mar 17, 2026
  b7d84534
- Fix deprecation: torch_dtype -> dtype · b49d3f59
  Your Name authored Mar 17, 2026
  
  b49d3f59
- Refactor: Move backend and manager classes to codai modules · 81c39eb8
  Your Name authored Mar 17, 2026
  81c39eb8
- Revert to working version from commit 001e1708 · 7c6b60f0
  Your Name authored Mar 17, 2026
  
  7c6b60f0
- Refactor: Move QueueManager to codai/queue/manager and restore FastAPI app · 989f1858
  Your Name authored Mar 17, 2026
  
  989f1858