ocr: lighton-ocr3-live.py, LightOnOCR-3 with a live view of the run (experimental)

maintenance/STATUS.md

@@ -3,7 +3,7 @@
 Generated by `python3 tools/recipe-status.py --write`. Do not edit by hand.
 Tiers and rules: [MAINTAINING.md](../MAINTAINING.md).
 
-Tiered: 1 of 52 recipes with a header.
+Tiered: 1 of 53 recipes with a header.
 
 | Recipe | Tier | Image | Packages | Last smoke test | Result |
 |---|---|---|---|---|---|
@@ -29,6 +29,7 @@ Tiered: 1 of 52 recipes with a header.
 | `ocr/lift-extract.py` | — | default | latest | never |  |
 | `ocr/lighton-ocr2-saturate.py` | — | `vllm/vllm-openai:v0.22.1` | latest | never |  |
 | `ocr/lighton-ocr3-bucket.py` | — | `vllm/vllm-openai:v0.30.0` | latest | never |  |
+| `ocr/lighton-ocr3-live.py` | — | `vllm/vllm-openai:v0.30.0` | latest | 2026-11-10 | [pass](https://huggingface.co/jobs/davanstrien/6af35d729737d9a9e0913810) |
 | `ocr/lighton-ocr3-saturate.py` | — | `vllm/vllm-openai:v0.30.0` | latest | never |  |
 | `ocr/lighton-ocr3.py` | — | `vllm/vllm-openai:v0.30.0` | latest | never |  |
 | `ocr/minicpm-ocr.py` | — | `cr.hf.co/davanstrien/minicpm-ocr:1.0-vllm-5d35ce3` | latest | never |  |

maintenance/verifications.jsonl

@@ -22,3 +22,4 @@
 {"date": "2026-11-10", "script": "embeddings/generate-embeddings-saturate.py", "commit": "32f5de9", "image": "vllm/vllm-openai:v0.30.0", "exclude_newer": "", "model_revision": "main", "flavor": "a10g-small", "input": "stanfordnlp/imdb (first 100 train rows)", "rows": 100, "outcome": "pass", "job_url": "https://huggingface.co/jobs/davanstrien/6af35312ab34849891ac6fcf", "notes": "--model BAAI/bge-m3 --max-seq-len 128: longest text 128 tokens, no prefix, 1024-d. 100/100 rows, 0 error rows, unit-length vectors, 100 distinct, 100 unique ids; card names the flavor, the prompt and the cut."}
 {"date": "2026-11-10", "script": "embeddings/generate-embeddings-saturate.py", "commit": "16457ba", "image": "vllm/vllm-openai:v0.30.0", "exclude_newer": "", "model_revision": "main", "flavor": "a10g-small", "input": "stanfordnlp/imdb (first 100 train rows)", "rows": 100, "outcome": "pass", "job_url": "https://huggingface.co/jobs/davanstrien/6af35597ab34849891ac74f2", "notes": "After the review fixes. Header only, default model Qwen3-Embedding-0.6B. 100/100 rows, 0 error rows, unit-length vectors, 100 unique ids; card has dim 1024 and the reproduce line carries --max-samples 100."}
 {"date": "2026-11-10", "script": "embeddings/generate-embeddings-saturate.py", "commit": "16457ba", "image": "vllm/vllm-openai:v0.30.0", "exclude_newer": "", "model_revision": "main", "flavor": "a10g-small", "input": "stanfordnlp/imdb (first 100 train rows)", "rows": 100, "outcome": "pass", "job_url": "https://huggingface.co/jobs/davanstrien/6af356829737d9a9e0912648", "notes": "Resume: the same command again on the same output. 0 rows processed, output still 100 rows with 100 unique ids; card keeps dim 1024 and says 100 rows were written by an earlier run."}
+{"date": "2026-11-10", "script": "ocr/lighton-ocr3-live.py", "commit": "0c07326", "image": "vllm/vllm-openai:v0.30.0", "exclude_newer": "", "model_revision": "main", "flavor": "rtx-pro-6000", "input": "davanstrien/encyclopaedia-britannica-1771 (first 60 rows)", "rows": 60, "outcome": "pass", "job_url": "https://huggingface.co/jobs/davanstrien/6af35d729737d9a9e0913810", "notes": "Header only, --live-port 7860 --expose-public 7860 --max-samples 60. 60/60 rows, 0 error rows, 60 unique ids, every finish_reason stop; 3 rows are blank leaves with empty text. Live view checked during the run: the page and ?view=single served, a page image served as JPEG, the feed counted 60 of 60 and reported finished."}

ocr/CLAUDE.md

@@ -88,6 +88,7 @@ Legend: ✅ production-ready · ⚠️ works only with a required pinned image 
 | `lighton-ocr2.py` | ✅ | vLLM | a100-large / l4x1 | resize 1540px; `--max-model-len` 16384 |
 | `lighton-ocr3-bucket.py` | ✅ | vLLM `:v0.30.0` | a10g-small | `hf://buckets/...` or dir → `.md` (+ `.json` blocks in grounding); bucket I/O via `bucketbag` v0.3.1 (no mount); `.jp2` via imagecodecs; resume-by-skip; failed pages write nothing |
 | `lighton-ocr3.py` | ✅ | vLLM `:v0.30.0` | a10g-small | 4B/0.8B Qwen3.5; resize 2048px; `--mode grounding` → `layout_blocks`; pinned upstream `lightonocr` client dep — see gotcha |
+| `lighton-ocr3-live.py` | 🧪 | saturate + vLLM `:v0.30.0` | rtx-pro-6000 | experimental: `lighton-ocr3-saturate.py` + `--live-port` page (finished pages at the run's pace; `?view=single`) |
 | `paddleocr-vl-1.5.py` | ✅ | transformers | l4x1 | not vLLM (server-only upstream); single-image |
 | `paddleocr-vl-1.6.py` | ✅ | vLLM | l4x1 | smart-resize ~1M px; SOTA OmniDocBench |
 | `paddleocr-vl.py` | ✅ | vLLM | l4x1 | |