host.today/ht-docker-ai
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(docker-images): add vLLM-based Nanonets-OCR2-3B image, Qwen3-VL Ollama image and refactor build/docs/tests to use new runtime/layout

Browse Source
This commit is contained in:
Jürgen Kunz
2026-01-19 21:05:51 +00:00
parent b58bcabc76
commit 08728ada4d
14 changed files with 1492 additions and 1126 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
changelog.md
View File

@@ -1,5 +1,16 @@
# Changelog
## 2026-01-19 - 1.14.0 - feat(docker-images)
add vLLM-based Nanonets-OCR2-3B image, Qwen3-VL Ollama image and refactor build/docs/tests to use new runtime/layout
- Add new Dockerfiles for Nanonets (Dockerfile_nanonets_vllm_gpu_VRAM10GB), Qwen3 (Dockerfile_qwen3vl_ollama_gpu_VRAM20GB) and a clarified MiniCPM Ollama variant (Dockerfile_minicpm45v_ollama_gpu_VRAM9GB); remove older, redundant Dockerfiles.
- Update build-images.sh to build the new image tags (minicpm45v, qwen3vl, nanonets-ocr) and adjust messaging/targets accordingly.
- Documentation overhaul: readme.md and readme.hints.md updated to reflect vLLM vs Ollama runtimes, corrected ports/VRAM estimates, volume recommendations, and API endpoint details.
- Tests updated to target the new model ID (nanonets/Nanonets-OCR2-3B), to process one page per batch, and to include a 10-minute AbortSignal timeout for OCR requests.
- Added focused extraction test suites (test/test.invoices.extraction.ts and test/test.invoices.failed.ts) for faster iteration and debugging of invoice extraction.
- Bump devDependencies: @git.zone/tsrun -> ^2.0.1 and @git.zone/tstest -> ^3.1.5.
- Misc: test helper references and docker compose/test port mapping fixed (nanonets uses 8000), and various README sections cleaned and reorganized.
## 2026-01-18 - 1.13.2 - fix(tests)
stabilize OCR extraction tests and manage GPU containers