Add resume-from-checkpoint: accumulate across preemptions

scripts/cloud/embed_modal.py

@@ -173,13 +173,33 @@ def embed_year(year: str = "2022", model_name: str = "ViT-L-14",
     def fetch_image(fig: dict) -> Image.Image | None:
         return None  # unused — we use scan_zip_and_embed instead
 
+    # ── Resume from existing HF checkpoint ───────────────────────────────────
+    # Download any existing embeddings from HF so we accumulate across preemptions
+    # rather than restarting from zero each time.
     all_ids: list[str] = []
     all_vecs: list[np.ndarray] = []
+    already_embedded: set[str] = set()
+
+    existing_path = f"/tmp/{year}_existing.parquet"
+    try:
+        existing_hf = hf_hub_download(
+            repo_id=OUT_REPO, filename=f"embeddings/{year}_vitl14.parquet",
+            repo_type="dataset", token=token, local_dir="/tmp/impact",
+        )
+        existing_df = pd.read_parquet(existing_hf)
+        already_embedded = set(existing_df["figure_id"].tolist())
+        # Seed all_ids / all_vecs with existing embeddings
+        all_ids = existing_df["figure_id"].tolist()
+        all_vecs = [existing_df["embedding"].to_numpy()]
+        print(f"Resuming from {len(already_embedded):,} existing embeddings on HF")
+    except Exception:
+        print("No existing checkpoint — starting fresh")
+
     batch_imgs: list[Image.Image] = []
     batch_ids: list[str] = []
     n_processed = 0
-    last_checkpoint_n = 0  # track last checkpoint boundary crossed
-    CHECKPOINT_EVERY = 25000
+    last_checkpoint_n = len(all_ids) // 10000  # start above existing checkpoints
+    CHECKPOINT_EVERY = 25000  # unused field — kept for reference
 
     def flush_batch():
         if not batch_imgs:
@@ -205,14 +225,20 @@ def embed_year(year: str = "2022", model_name: str = "ViT-L-14",
             continue
 
         pid_norm = str(pid_raw).lstrip("D").zfill(7)
-        batch_ids.append(f"D{pid_norm}_{fig_num}")
+        fig_id = f"D{pid_norm}_{fig_num}"
+
+        # Skip already-embedded figures
+        if fig_id in already_embedded:
+            continue
+
+        batch_ids.append(fig_id)
         batch_imgs.append(img)
         if len(batch_imgs) >= batch_size:
             flush_batch()
 
-        # Checkpoint every 10k to HF — accumulates progress across preemptions
+        # Checkpoint every 10k NEW embeddings to HF
         current_boundary = len(all_ids) // 10000
-        if current_boundary > last_checkpoint_n:
+        if current_boundary > last_checkpoint_n and len(all_ids) > len(already_embedded):
             _vecs_cp = np.vstack(all_vecs).astype(np.float32)
             _norms_cp = np.linalg.norm(_vecs_cp, axis=1, keepdims=True)
             _vecs_cp /= np.maximum(_norms_cp, 1e-8)