runpod: a pod can be told to return token ids (v0.3.32)

`return_token_ids` added to the vllm entry of _ENGINE_FORWARDED_FIELDS, so a
rented pod's vLLM backend can be asked for raw vocabulary ids -- without which
the v0.3.31 repair can never fire on a RunPod-served model, which is every OCR
model on the Digesta install.

Inert by default: engine_config_for_pod skips falsey values, so adding the field
cannot change an install that has not set it.

Set on Aruba for `datalab-to/surya-ocr-2` ONLY, through its models.json `vllm`
block rather than globally. The flag changes what the `token` string inside
logprobs looks like and that install also serves qwen38-awq and diogene-alpha on
vLLM pods; nothing in coderai reads that field except the OCR repair, but a
client might, and there is no reason to find out the hard way.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01PxYHWbCAjjjD7A1rhFKtaW

codai/__init__.py

@@ -16,7 +16,7 @@
 
 # Canonical product version for CoderAI — single source of truth. Both the API
 # metadata and the admin web UI read from here.
-__version__ = "0.3.31"
+__version__ = "0.3.32"
 __tagline__ = "Complete orchestration, distribution and escalation of remotizable advanced inference."
 
 # Configure the CUDA caching allocator BEFORE torch is imported anywhere.

codai/api/runpod_worker.py

@@ -919,7 +919,12 @@ _ENGINE_FORWARDED_FIELDS = {
            "model_id"),
     "kt": ("ctx", "extra_args", "extra_env", "model_id"),
     "vllm": ("ctx", "gpu_memory_utilization", "tensor_parallel_size", "max_num_seqs",
-             "dtype", "quantization", "extra_args", "extra_env"),
+             "dtype", "quantization", "extra_args", "extra_env",
+             # Only ever forwarded when TRUE: the loop below skips False, so
+             # adding it here cannot change an install that has not asked for it.
+             # A pod needs it to recover a byte-level vocabulary's characters from
+             # the token ids (see VllmConfig.return_token_ids).
+             "return_token_ids"),
 }
 #: The config attribute behind each engine name (ktransformers is the odd one).
 _ENGINE_CONFIG_ATTR = {"kt": "ktransformers"}

tests/test_ocr_accent_repair.py

@@ -404,3 +404,19 @@ def test_it_can_be_set_per_model():
     changing what every other vLLM model reports."""
     from codai.backends.overrides import VLLM_FIELDS
     assert "return_token_ids" in VLLM_FIELDS
+
+
+def test_a_pod_can_be_told_to_return_token_ids():
+    """The pod's vLLM backend is configured by what this install forwards, so the
+    field has to be in the forwarded set or a pod could never ask for ids."""
+    from codai.api.runpod_worker import _ENGINE_FORWARDED_FIELDS
+    assert "return_token_ids" in _ENGINE_FORWARDED_FIELDS["vllm"]
+
+
+def test_forwarding_it_cannot_affect_an_install_that_did_not_ask():
+    """engine_config_for_pod skips falsey values, so adding the field to the
+    forwarded set is inert until someone sets it True."""
+    import inspect
+    from codai.api import runpod_worker
+    src = inspect.getsource(runpod_worker.engine_config_for_pod)
+    assert "if v in (None, \"\", 0, False):" in src