Expose llama.cpp worker errors

Co-authored-by: Codex <chatgpt-codex-connector[bot]@users.noreply.github.com>

src/extraction/llamacpp_gpu.py

@@ -126,7 +126,14 @@ def _run_llamacpp_generation(
     chat_handler: str,
     max_tokens: int,
 ) -> str:
-    model_path, mmproj_path = _download(repo, model_file, mmproj_file)
+    try:
+        model_path, mmproj_path = _download(repo, model_file, mmproj_file)
+    except Exception as exc:
+        raise RuntimeError(
+            "llama.cpp download failed while preparing the GGUF/mmproj pair: "
+            f"{type(exc).__name__}: {exc}"
+        ) from exc
+
     try:
         llm = _load(model_path, mmproj_path, chat_handler)
     except Exception as exc:
@@ -134,13 +141,21 @@ def _run_llamacpp_generation(
             "The llama.cpp backend could not load the MiniCPM-V GGUF/mmproj pair. "
             "This usually means the downloaded model build is incompatible with the installed "
             "llama-cpp-python wheel, the chat handler name does not match the wheel, or one of "
-            "the model files is incomplete."
+            "the model files is incomplete. "
+            f"Inner error: {type(exc).__name__}: {exc}"
+        ) from exc
+
+    try:
+        response = llm.create_chat_completion(
+            messages=[{"role": "user", "content": [{"type": "text", "text": EXTRACTION_PROMPT}, *parts]}],
+            response_format={"type": "json_object"},
+            temperature=0.0,
+            reasoning_budget=0,
+            max_tokens=max_tokens,
+        )
+        return response["choices"][0]["message"].get("content") or "{}"
+    except Exception as exc:
+        raise RuntimeError(
+            "llama.cpp generation failed while extracting the document. "
+            f"Inner error: {type(exc).__name__}: {exc}"
         ) from exc
-    response = llm.create_chat_completion(
-        messages=[{"role": "user", "content": [{"type": "text", "text": EXTRACTION_PROMPT}, *parts]}],
-        response_format={"type": "json_object"},
-        temperature=0.0,
-        reasoning_budget=0,
-        max_tokens=max_tokens,
-    )
-    return response["choices"][0]["message"].get("content") or "{}"