From e8d3dd44b37241be35723143404ef8c8a6188756 Mon Sep 17 00:00:00 2001 From: Rafael Poyiadzi Date: Thu, 12 Feb 2026 15:08:55 +0000 Subject: [PATCH] Fix docs images not rendering Images in docs/images/ weren't included in the Next.js static export. Copied them to docs-site/public/images/ so they're included in the build output, and updated markdown to use absolute paths with the /docs basePath. Co-Authored-By: Claude Opus 4.6 --- docs-site/public/images/fda_10pct_scaling.png | 3 +++ docs-site/public/images/learning_curve_accuracy.png | 3 +++ docs/active-learning-llm-oracle.md | 2 +- docs/scale-deduplication-20k-rows.md | 2 +- 4 files changed, 8 insertions(+), 2 deletions(-) create mode 100644 docs-site/public/images/fda_10pct_scaling.png create mode 100644 docs-site/public/images/learning_curve_accuracy.png diff --git a/docs-site/public/images/fda_10pct_scaling.png b/docs-site/public/images/fda_10pct_scaling.png new file mode 100644 index 00000000..b38f5d01 --- /dev/null +++ b/docs-site/public/images/fda_10pct_scaling.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b2b4b0dc668c3051ed25e7cec0f956fb0d376a059e0620987ac17468c1271fcc +size 326137 diff --git a/docs-site/public/images/learning_curve_accuracy.png b/docs-site/public/images/learning_curve_accuracy.png new file mode 100644 index 00000000..4a15e184 --- /dev/null +++ b/docs-site/public/images/learning_curve_accuracy.png @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bb8f6f2cadf3d8c2718c77576268862075ad4f5f0d5ec82ed521f66224706143 +size 116405 diff --git a/docs/active-learning-llm-oracle.md b/docs/active-learning-llm-oracle.md index c2de146e..7bd3ac33 100644 --- a/docs/active-learning-llm-oracle.md +++ b/docs/active-learning-llm-oracle.md @@ -5,7 +5,7 @@ description: Use everyrow's agent_map as an LLM oracle in an active learning loo # How to Replace Human Data Annotators with LLMs in Active Learning -![Active Learning: Ground Truth vs LLM Oracle](images/learning_curve_accuracy.png) +![Active Learning: Ground Truth vs LLM Oracle](/docs/images/learning_curve_accuracy.png) Human data labeling is slow and expensive. We replaced the human annotator with an LLM oracle in an active learning loop and achieved identical classifier performance — 200 labels in under 5 minutes for $0.26. diff --git a/docs/scale-deduplication-20k-rows.md b/docs/scale-deduplication-20k-rows.md index 7a264c6a..8e813451 100644 --- a/docs/scale-deduplication-20k-rows.md +++ b/docs/scale-deduplication-20k-rows.md @@ -7,7 +7,7 @@ description: Scale LLM-powered deduplication to 20,000 rows with linear cost, ac LLM-powered deduplication gives you semantic understanding that string matching can't, but naive pairwise comparison is quadratic. At 20,000 rows that's 200 million pairs. Everyrow's dedupe pipeline uses a funnel of embeddings, clustering, and targeted LLM calls to keep cost linear and accuracy high. -![FDA Drug Products — Deduplication at Scale](images/fda_10pct_scaling.png) +![FDA Drug Products — Deduplication at Scale](/docs/images/fda_10pct_scaling.png) Error rates stay near zero as scale increases. Cost and LLM calls scale linearly. Runtime is under 5 minutes up to 10,000 rows and 25 minutes at 20,000.