Stephanos Pipeline Progress

Last updated: 2026-07-12 22:14:50

Stage Done Total Progress % Rate ETA
AI Translation 311 3,569
8.7% 274/week 2.8 months
Human Review 202 311
65.0% 0/week stalled
Meineke Page OCR
Queued missing-page OCR for the public Meineke source text
 835 835
100.0% 0/week complete
Proper Noun Extraction 312 312
100.0% 0/week complete
Structured Source Citations
3,087 current texts with citations; 483 completed with zero units; 0 failed without a later completed run; 831,875 tokens recorded, 0 in the last 7 days
 3,570 3,570
100.0% 0/week complete
Wikidata Sources 432 432
100.0% 0/week complete
Etymology Extraction 312 312
100.0% 0/week complete
Alias Extraction 1,364 1,364
100.0% 0/week complete
Spelling Variants 9,542 9,542
100.0% 0/week complete
Human Entity Review 4,158 9,697
42.9% 0/week stalled
Wikidata Places 2,875 3,084
93.2% 60/week 3.5 weeks
Meineke Difference Analysis 2,001 2,001
100.0% 0/week complete
Text Pair Sync 3,569 3,569
100.0% 3,569/week complete
Translation Risk Sync 311 311
100.0% 312/week complete
Prompt Guidance Checks
3,570 current Meineke headwords x 101 prompt guidance rules (formula, gloss, contextual_bias; translation_guidance_scan_v4); default nightly limit 2,000 checks; queue pending/running/completed/failed/cancelled: 78,895/0/43,873/1/0; 26,850,851 guidance AI tokens recorded, 803,323 in the last 7 days
 43,820 360,570
12.2% 1,014/week 5.3 months
Stylometric Fingerprinting
Broad UMAP-ready formula vectors require 21/32 active formula rules: 2,808 entries (320 Kappa, 13 Parisinus); complete formula coverage: 380 entries (1 Parisinus); see statistics/fingerprinting.html; sentence grammar rows: sentence_grammar_runs=8, sentence_grammar_evaluations=111, sentence_grammar_tokens=495
 2,808 3,570
78.7% 524/week 1.5 weeks
German Reference Page OCR
Internal reference lane only; statuses entries/no-headword/not-German/other: 161/9/11/21; default nightly limit 3 pages; 1,080,410 OCR tokens recorded
 202 806
25.1% 18/week 6.7 months
German Reference Translation
Internal reference lane only; not public and not part of the main translation queue; failed: 0; default nightly limit 5 refs; 187,881 tokens recorded, 15,916 in the last 7 days
 312 347
89.9% 29/week 1.0 weeks
Meineke Word Lemmatization
status pending/processing/error: 0/0/0; 12,799,426 tokens recorded, 0 in the last 7 days
 3,569 3,569
100.0% 2/week complete
Paper Translation Model Coverage
25 tracked paper/evaluation model variants; Responses worker limit 1 generated run(s) per pipeline run; experiment enqueue limit 1 new request(s) per profile per run
 1,446 2,900
49.9% 309/week see model table
nodegoat Sync 3,560 3,569
99.7% 0/week stalled

Paper Translation Model Coverage

This table uses the same 100-row Kappa paper corpus as the prompt-evaluation page. Claude rows are external imports; OpenAI rows are generated by the daily pipeline.

Model / prompt API / reasoning Kappa rows Runs Open Runs 7d Latest Projection
claude_fable_5 v1 chat_completions 100/100 100/100 0 0 2026-07-05 complete
claude_fable_5 v2 chat_completions 100/100 100/100 0 0 2026-07-05 complete
claude_fable_5 v3 chat_completions 100/100 100/100 0 0 2026-07-03 complete
claude_opus_4_8 v1 chat_completions 100/100 100/100 0 0 2026-07-05 complete
claude_opus_4_8 v2 chat_completions 0/100 0/100 0 0 N/A external Claude import required; not scheduled in this pipeline
claude_opus_4_8 v3 chat_completions 100/100 100/100 0 0 2026-07-05 complete
claude_sonnet_5 v1 chat_completions 100/100 100/100 0 0 2026-07-03 complete
claude_sonnet_5 v2 chat_completions 100/100 100/100 0 0 2026-07-03 complete
claude_sonnet_5 v3 chat_completions 100/100 100/100 0 0 2026-07-05 complete
gpt-5.4 v1 chat_completions 0/100 0/100 0 0 N/A no open request; waits for next enqueue pass
gpt-5.4 v2 chat_completions 0/100 0/100 0 0 N/A no open request; waits for next enqueue pass
gpt-5.4 v3 chat_completions 0/100 0/100 0 0 N/A no open request; waits for next enqueue pass
gpt-5.5 v1 chat_completions 100/100 104/100 0 0 2026-06-21 complete
gpt-5.5 v2 chat_completions 100/100 100/100 0 0 2026-06-21 complete
gpt-5.5 v3 chat_completions 100/100 116/100 0 0 2026-06-20 complete
gpt-5.5_v3_repeat v1 chat_completions 6/100 30/500 0 5 2026-07-07 about 3.1 months at one daily run
gpt-5.5_v4_mini v1 responses / low 100/100 100/100 0 98 2026-07-08 complete
gpt-5.5_v4_mini_high v1 responses / high 100/100 100/100 0 98 2026-07-08 complete
gpt-5.5_v4_mini_medium v1 responses / medium 100/100 100/100 0 98 2026-07-08 complete
gpt-5.5_v4_reasoning v1 responses / low 12/100 12/100 0 10 2026-07-12 shared Responses lane: 284 generated runs remain across 6 profiles; about 9.5 months at one daily run at TRANSLATION_RESPONSES_RUN_LIMIT=1
gpt-5.5_v4_reasoning_high v1 responses / high 2/100 2/100 10 0 2026-07-01 shared Responses lane: 284 generated runs remain across 6 profiles; about 9.5 months at one daily run at TRANSLATION_RESPONSES_RUN_LIMIT=1
gpt-5.5_v4_reasoning_medium v1 responses / medium 2/100 2/100 10 0 2026-06-30 shared Responses lane: 284 generated runs remain across 6 profiles; about 9.5 months at one daily run at TRANSLATION_RESPONSES_RUN_LIMIT=1
gpt-5.6-sol v1 chat_completions 0/100 0/100 0 0 N/A no open request; waits for next enqueue pass
gpt-5.6-sol v2 chat_completions 0/100 0/100 0 0 N/A no open request; waits for next enqueue pass
gpt-5.6-sol v3 chat_completions 0/100 0/100 0 0 N/A no open request; waits for next enqueue pass
Note: ETA estimates use the faster of the observed 7-day rate and any configured nightly cron limit known to this page. "Stalled" means no observed progress and no configured nightly rate is known. Proper-noun and etymology extraction are measured over translated lemmas, matching the nightly jobs. Wikidata stages count rows once they have a recorded outcome, including not-found, ambiguous, and human-reviewed cases. Retired legacy Greek OCR stages are intentionally omitted; the active German-reference lane is internal and reported separately when present.