renderers 0.1.13.dev5__tar.gz → 0.1.13.dev7__tar.gz

{renderers-0.1.13.dev5 → renderers-0.1.13.dev7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: renderers
-Version: 0.1.13.dev5
+Version: 0.1.13.dev7
 Summary: Chat template renderers — deterministic message-to-token conversion for LLM training
 License-Expression: Apache-2.0
 License-File: LICENSE

{renderers-0.1.13.dev5 → renderers-0.1.13.dev7}/renderers/_version.py

@@ -18,7 +18,7 @@ version_tuple: tuple[int | str, ...]
 commit_id: str | None
 __commit_id__: str | None
 
-__version__ = version = '0.1.13.dev5'
-__version_tuple__ = version_tuple = (0, 1, 13, 'dev5')
+__version__ = version = '0.1.13.dev7'
+__version_tuple__ = version_tuple = (0, 1, 13, 'dev7')
 
 __commit_id__ = commit_id = None

{renderers-0.1.13.dev5 → renderers-0.1.13.dev7}/renderers/client.py

@@ -261,9 +261,11 @@ async def generate(
     Returns a dict with: request_id, prompt_ids, renderer_prompt_ids,
     mm_placeholders, completion_ids, completion_logprobs, content,
     reasoning_content, tool_calls, finish_reason, routed_experts,
-    multi_modal_data, prompt_attribution. ``renderer_prompt_ids`` is the
-    unexpanded logical prompt when ``process_multimodal=False`` and ``None``
-    otherwise.
+    weight_versions, multi_modal_data, prompt_attribution.
+    ``renderer_prompt_ids`` is the unexpanded logical prompt when
+    ``process_multimodal=False`` and ``None`` otherwise. ``weight_versions``
+    is vLLM's (>= 0.32) list of ``{version, start, end}`` spans over
+    completion-token indexes, or ``None`` when the engine labels none.
 
     ``prompt_attribution`` is the renderer's :class:`RenderedTokens` for
     the prompt — either the one this call computed via
@@ -431,6 +433,7 @@ async def generate(
         "reasoning_complete": parsed.reasoning_complete,
         "routed_experts": routed_experts,
         "sampling_mask": sampling_mask,
+        "weight_versions": choice.get("weight_versions"),
         # The mm sidecar consumed on the request side, surfaced back so
         # callers can persist it on the trajectory step for downstream
         # multi-turn bridging and training-sample construction.