Upload tools/run_embedding_job.py

Upload tools/run_embedding_job.py

Upload tools/build_product_embeddings.py

Upload tools/build_product_embeddings.py

Upload tools/run_embedding_job.py

Upload tools/run_embedding_job.py

fix: use CLIP ViT-B/32 vision tower to match the app's ViT-B/32-aligned multilingual text tower (was wrongly patch16); no translation model needed

feat: end-to-end HF Job entrypoint for product embeddings

fix: mirror embedding outputs to dist/ so the browser can fetch them; add numpy check

Upload tools/build_product_embeddings.py