This website requires JavaScript.
c1dae49ba7
fix(render): two silent hangs in the turbo capture path (#132 )
2026-07-31 16:51:58 +08:00
ae809153b5
fix(render): measure page height from content, not the body box (#131 )
2026-07-31 16:43:37 +08:00
aa8b7646f2
fix(render): measure page height from content, not the body box
2026-07-30 01:49:41 +08:00
7a3c3041bd
feat(serve): distinguish search vs chat in query log via X-Source header (#127 )
2026-07-27 15:54:41 -07:00
6d9bbf0abb
feat(serve): distinguish search vs chat in query log via X-Source header
2026-07-27 15:54:23 -07:00
d745db7b23
chore: fix CI lint for ML codebase (#126 )
2026-07-27 15:53:00 -07:00
1df51cd492
chore: remove accidentally staged video and gitignore mp4s
2026-07-27 15:04:53 -07:00
60afed71d5
chore: fix CI lint — sort imports and relax ruff rules for ML codebase
2026-07-27 15:04:47 -07:00
1bfbb5d215
feat(serve): write search request metadata to local JSONL (#125 )
2026-07-27 14:14:26 -07:00
2b7f055f25
feat(serve): write search request metadata to local JSONL
2026-07-27 14:14:06 -07:00
1a9ba75620
feat: Vector search with Qdrant (#111 )
2026-07-16 16:34:42 +05:30
05a8f1f7c1
feat(cli): add batch URL input support (#64 )
2026-07-16 19:03:19 +08:00
03369a7809
chore: bump version to 0.4.0 (#121 )
2026-07-16 18:43:27 +08:00
8ff8f1fa8b
chore: bump version to 0.4.0
2026-07-16 18:41:54 +08:00
1ee3d8e18d
feat(render): networkidle2 in-flight wait, replacing the completion-reset probe (#120 )
2026-07-16 18:36:15 +08:00
f7cb811d7f
feat(render): networkidle2 in-flight wait, replacing the completion-reset probe
2026-07-16 18:34:56 +08:00
bfe11a3199
fix: post-merge audit follow-ups (idle-wait default, incremental re-runs, eval config, progress, alpha) (#119 )
2026-07-16 18:17:03 +08:00
b8b5fa4d74
fix(index): composite transparent local images onto white, not black
2026-07-16 18:15:17 +08:00
45810a912e
fix(embed): keep chunk progress bar out of pool workers; restore loggable progress
2026-07-16 18:14:49 +08:00
6514410eb5
fix(eval): let the model registry's api_base/api_key actually take effect
2026-07-16 18:14:13 +08:00
498399afaa
fix(index): detect stale tile dirs on incremental re-runs; harden manifest writes
2026-07-16 18:13:32 +08:00
cee72c3011
fix(index): make wait_network_idle default source-aware, not blanket
2026-07-16 18:10:25 +08:00
cbb850d26e
fix(cli): accept auto/mps devices in pixelrag index build (#73 )
2026-07-16 02:34:15 -07:00
552ecfd299
feat(index,serve): department-filtered search for local document indexes (#115 )
2026-07-16 16:30:05 +07:00
71083a2714
feat(eval): add MiniMax model configs (#113 )
2026-07-16 17:28:59 +08:00
2a005dbc4e
feat(embed): add tqdm progress bar to chunk stage, clean up embed logging (#108 )
2026-07-16 02:23:15 -07:00
9d84f021b4
fix(index): write article_id into tile manifests, stop guessing from dir names (#83 )
2026-07-16 17:19:53 +08:00
20b88696fa
train/sft: add type annotations to all Python scripts (#85 )
2026-07-16 14:49:48 +05:30
66a070176e
Merge remote-tracking branch 'origin/main' into pr83
2026-07-16 17:16:26 +08:00
75be53abcb
Merge branch 'main' into feat/progress-bars
2026-07-16 17:10:21 +08:00
3941e72fd7
fix(chrome): sort Playwright Chromium by revision, not lexically (#74 )
2026-07-16 02:09:44 -07:00
4317dc9f6f
chore(eval): make the reproduction package self-contained and public-safe (#90 )
2026-07-16 17:07:11 +08:00
37b7692050
Merge remote-tracking branch 'origin/main' into pr90
2026-07-16 17:05:22 +08:00
72fab8873d
fix(eval): sanitize example_id to prevent path traversal (fixes #78 , #79 ) (#80 )
2026-07-16 14:33:17 +05:30
fc7f430fcf
feat(index): render local images to tile directories (#95 )
2026-07-16 14:33:12 +05:30
2b7ea1a945
feat(index): default to --wait-network-idle for URL rendering (#96 )
2026-07-16 14:33:08 +05:30
7af960e745
fix(deps): move tqdm, pyyaml, numpy from heavy extras into core (#118 )
2026-07-16 16:57:07 +08:00
2d0ff8aaed
docs: add citation section and link to arxiv paper (#112 )
2026-06-29 20:40:11 -07:00
b97c0ba962
docs: add citation section and link to arxiv paper
2026-06-29 20:39:57 -07:00
3b6f7af360
feat(embed): add tqdm progress bar to chunk stage, clean up embed logging
2026-06-27 06:12:00 -07:00
83ccea4477
fix(index): render markdown properly and resolve relative image paths (#107 )
2026-06-27 01:03:26 -07:00
d84720178b
style: ruff format test_local_source_text.py
2026-06-27 00:58:05 -07:00
ebbdf53db0
style: ruff format pipelines.py
2026-06-27 00:55:33 -07:00
9e75730e0f
fix(index): render markdown properly and resolve relative image paths
2026-06-27 00:52:55 -07:00
55ea7c5594
feat(index): support .md and .txt files natively in the pipeline (#101 )
2026-06-27 13:19:57 +05:30
1c23d8a31f
perf(serve): optional mmap index load (PIXELRAG_INDEX_MMAP=1) for near-instant startup
2026-06-26 11:21:41 -07:00
1727ea8974
perf(serve): reuse a persistent Chrome for on-demand render (--cdp-url), ~3x faster
2026-06-26 10:42:11 -07:00
a8cad07455
docs(render): note throughput is capture-only; measure with the bench harness
2026-06-26 10:04:06 -07:00
c3eda5aaf5
fix(render): default GPU rasterization OFF (it was an inherited no-op that crashes some boxes)
2026-06-26 08:56:46 -07:00
d8bf1f93b7
refactor(serve): render on-demand tiles via the pixelshot CLI, not inline python -c
2026-06-26 08:49:26 -07:00
dd21843c70
fix(serve): render on-demand tiles in a subprocess so the serve doesn't wedge
2026-06-26 08:31:04 -07:00
0cf430ad85
fix(eval): configurable retrieval timeout so a slow serve doesn't silently go closed-book
2026-06-26 05:41:08 -07:00
4879a13073
revert(render): keep the original GPU-rasterization flags (no-GPU boxes render fine)
2026-06-26 05:39:18 -07:00
a770ad7dfc
fix(render): default to CPU rasterization (the render machines have no GPU)
2026-06-26 04:00:31 -07:00
41bfb3765c
fix(serve,render): make Mode-3 on-demand rendering actually work
2026-06-26 03:36:02 -07:00
f9c29a5e30
fix(serve)+docs(eval): add torchvision to the serve extra; document self-hosting the search serve
2026-06-25 20:53:18 -07:00
50396c21b6
fix(eval): grade NQ/NQ-Tables with the LLM judge to match the paper
2026-06-25 20:45:52 -07:00
879e7367ec
chore(eval): relax deps to floors + optional reader extra; document the public-API path
2026-06-24 02:01:06 -07:00
53a4a3b019
fix(eval): grade NQ correctly and feed serve-returned base64 tiles to the reader
2026-06-24 01:06:59 -07:00
1d151d0503
chore(eval): make the reproduction package self-contained and public-safe
2026-06-23 22:16:58 -07:00
ddd2589d1b
fix(index): make article_id manifest contract actually reach the GPU path
2026-06-23 20:36:25 -07:00
28c9b5c420
fix(index): write article_id into tile manifests, stop guessing from dir names
2026-06-23 06:00:25 -07:00
13d53c1250
chore: remove uptime monitoring from main repo (#82 )
2026-06-23 18:19:16 +08:00
09508b9395
🗃️ Update status summary [skip ci] [upptime]
2026-06-23 10:06:46 +00:00
ea8de93ebb
chore: remove uptime monitoring from main repo
2026-06-23 02:52:23 -07:00
b4ff2b26a1
Delete CNAME
2026-06-23 02:40:04 -07:00
2ac42b773d
📊 Update status summary [skip ci]
2026-06-23 09:13:53 +00:00
2516bcfe0d
fix(status): replace Upptime readme with custom summary generator
2026-06-23 02:13:17 -07:00
8d5c607f9e
feat(cdp): --cdp-url to attach to an already-running browser (supersedes #55 ) (#76 )
2026-06-23 11:48:06 +08:00
9cc1feb798
fix: handle non-numeric filename-stem IDs in local source pipeline (#66 )
2026-06-23 05:22:22 +02:00
6184b0dc96
fix: add timeouts, improve UX (back button, lightbox in chat), prevent agent loops (#49 )
2026-06-23 11:22:18 +08:00
f71d1b1000
fix(render): isolate Chrome --user-data-dir per worker (fixes #54 ) (#65 )
2026-06-23 08:50:54 +05:30
5756fe7c12
chore: add pre-commit hook (ruff format + lint) and bump to 0.3.0 (#75 )
2026-06-23 11:18:09 +08:00
fad94c82c9
🟩 Website is up (200 in 119 ms) [skip ci] [upptime]
2026-06-23 00:04:27 +00:00
53f1afecda
🟩 Agent is up (200 in 97 ms) [skip ci] [upptime]
2026-06-23 00:04:27 +00:00
944038db69
🟩 Tile Serving is up (200 in 156 ms) [skip ci] [upptime]
2026-06-23 00:04:27 +00:00
e700053963
🟩 Search API is up (200 in 97 ms) [skip ci] [upptime]
2026-06-23 00:04:27 +00:00
3a9bb467bf
🟩 API Health is up (200 in 97 ms) [skip ci] [upptime]
2026-06-23 00:04:27 +00:00
096adf537d
fix(embed): 2D-tile wide pages at native resolution (PDFs/landscape) (#48 )
2026-06-22 22:56:09 +08:00
ac918d9175
Deploying to gh-pages from @ StarTrail-org/PixelRAG@bfe6917ea5 🚀
2026-06-22 14:31:47 +00:00
bfe6917ea5
chore(uptime): write monitoring data to a dedicated status branch, not main (#68 )
2026-06-22 22:31:29 +08:00
e83564e74d
🗃️ Update status summary [skip ci] [upptime]
2026-06-22 14:30:45 +00:00
6da67820bf
docs(pixelbrowse): recommend uv tool/pipx install so pixelshot is on PATH (#45 )
2026-06-22 22:20:41 +08:00
22b46d648d
docs: add pixelshot CLI usage examples (#62 )
2026-06-21 18:21:15 -07:00
8fd072c15e
docs: move CLI examples below programmatic section
2026-06-21 18:21:09 -07:00
5b6f908c60
docs: use real tested URLs in pixelshot CLI examples
2026-06-21 18:16:23 -07:00
788d8b238e
docs: add pixelshot CLI usage examples to README
2026-06-21 18:11:06 -07:00
da4060ae0f
feat: add macOS support to uv lockfile and torch sources (#61 )
2026-06-21 18:10:14 -07:00
7800c55e82
feat: add macOS support to uv lockfile and torch sources
2026-06-21 18:08:10 -07:00
1119c84e2e
Fix CDP capture hang when Chrome exposes component-extension targets (#56 )
2026-06-22 07:26:59 +07:00
6d9cdda76f
docs(train): link synthetic data pipeline from Data Curation section (#60 )
2026-06-21 16:18:32 -07:00
743a2ff7ba
docs: move Data Curation section from root README to train/README (#59 )
2026-06-21 16:03:19 -07:00
23e6b3b722
docs: move Data Curation section from root README to train/README
2026-06-21 16:02:33 -07:00
86e1de4f41
docs: add collapsible PDF quickstart example to README (#58 )
2026-06-21 15:55:07 -07:00
be0a941282
fix(embed): resize wide PDF chunks instead of skipping them (#51 )
2026-06-21 15:45:23 -07:00
e93f8d5f5e
feat(api): add request ID middleware for log tracing (#41 )
2026-06-13 01:57:12 +08:00
7ba3895195
feat(chrome): cross-platform Chrome discovery (macOS/Windows, not just Linux) (#44 )
2026-06-12 16:33:26 +08:00
728af8ff96
feat(chrome): cross-platform Chrome discovery (macOS/Windows, not just Linux)
2026-06-12 01:31:31 -07:00
3e85fcb131
fix(render): cap requestAnimationFrame waits in the readiness + scroll probes (#39 )
2026-06-11 19:59:00 +08:00
fca912ecdd
fix(render): cap requestAnimationFrame waits in the readiness + scroll probes
2026-06-11 04:57:27 -07:00
Zhifei Li