[overlap-spec] fix stop condition and trimming (#11819)

python/sglang/srt/managers/detokenizer_manager.py

@@ -142,6 +142,7 @@ class DetokenizerManager(MultiHttpWorkerDetokenizerMixin):
             if output[-1] == 200012 and self.is_tool_call_parser_gpt_oss:
                 return output
             assert len(output) > 0
+            # NOTE: We can always assume the last token is the matched stop token
             return output[:-1]
         return output
 

python/sglang/srt/managers/schedule_batch.py

@@ -486,6 +486,8 @@ class Req:
         # Check finish
         self.tokenizer = None
         self.finished_reason = None
+        # finished position (in output_ids), used when checking stop conditions with speculative decoding
+        self.finished_len = None
         # Whether this request has finished output
         self.finished_output = None
         # If we want to abort the request in the middle of the event loop, set this to true
@@ -651,6 +653,13 @@ class Req:
         spec_alg = get_global_server_args().speculative_algorithm
         return self.sampling_params.max_new_tokens == 0 and spec_alg is None
 
+    @property
+    def output_ids_through_stop(self) -> List[int]:
+        """Get the output ids through the stop condition. Stop position is included."""
+        if self.finished_len is not None:
+            return self.output_ids[: self.finished_len]
+        return self.output_ids
+
     def add_latency(self, stage: RequestStage):
         if self.metrics_collector is None:
             return
@@ -702,18 +711,20 @@ class Req:
     def init_incremental_detokenize(self):
         first_iter = self.surr_offset is None or self.read_offset is None
 
+        output_ids = self.output_ids_through_stop
+
         if first_iter:
             self.read_offset = len(self.origin_input_ids_unpadded)
             self.surr_offset = max(
                 self.read_offset - INIT_INCREMENTAL_DETOKENIZATION_OFFSET, 0
             )
             self.surr_and_decode_ids = (
-                self.origin_input_ids_unpadded[self.surr_offset :] + self.output_ids
+                self.origin_input_ids_unpadded[self.surr_offset :] + output_ids
             )
-            self.cur_decode_ids_len = len(self.output_ids)
+            self.cur_decode_ids_len = len(output_ids)
         else:
-            self.surr_and_decode_ids.extend(self.output_ids[self.cur_decode_ids_len :])
-            self.cur_decode_ids_len = len(self.output_ids)
+            self.surr_and_decode_ids.extend(output_ids[self.cur_decode_ids_len :])
+            self.cur_decode_ids_len = len(output_ids)
 
         return self.surr_and_decode_ids, self.read_offset - self.surr_offset
 
@@ -760,55 +771,31 @@ class Req:
 
         return False
 
-    def check_finished(self):
-        if self.finished():
-            return
-
-        if self.to_abort:
-            self.finished_reason = FINISH_ABORT(
-                message=self.to_abort_message,
-            )
-            return
-
-        if len(self.output_ids) >= self.sampling_params.max_new_tokens:
-            self.finished_reason = FINISH_LENGTH(
-                length=self.sampling_params.max_new_tokens
-            )
-            return
-
-        if self.grammar is not None:
-            if self.grammar.is_terminated():
-                self.finished_reason = FINISH_MATCHED_TOKEN(matched=self.output_ids[-1])
-                return
-
-        last_token_id = self.output_ids[-1]
+    def _check_token_based_finish(self, new_accepted_tokens: List[int]) -> bool:
+        if self.sampling_params.ignore_eos:
+            return False
 
-        if not self.sampling_params.ignore_eos:
+        # Check stop token ids
         matched_eos = False
 
-            # Check stop token ids
+        for i, token_id in enumerate(new_accepted_tokens):
             if self.sampling_params.stop_token_ids:
-                matched_eos = last_token_id in self.sampling_params.stop_token_ids
+                matched_eos |= token_id in self.sampling_params.stop_token_ids
             if self.eos_token_ids:
-                matched_eos |= last_token_id in self.eos_token_ids
+                matched_eos |= token_id in self.eos_token_ids
             if self.tokenizer is not None:
-                matched_eos |= last_token_id == self.tokenizer.eos_token_id
+                matched_eos |= token_id == self.tokenizer.eos_token_id
                 if self.tokenizer.additional_stop_token_ids:
-                    matched_eos |= (
-                        last_token_id in self.tokenizer.additional_stop_token_ids
-                    )
+                    matched_eos |= token_id in self.tokenizer.additional_stop_token_ids
             if matched_eos:
-                self.finished_reason = FINISH_MATCHED_TOKEN(matched=last_token_id)
-                return
+                self.finished_reason = FINISH_MATCHED_TOKEN(matched=token_id)
+                matched_pos = len(self.output_ids) - len(new_accepted_tokens) + i
+                self.finished_len = matched_pos + 1
+                return True
 
-        if last_token_id > self.vocab_size or last_token_id < 0:
-            if self.sampling_params.stop_token_ids:
-                self.output_ids[-1] = next(iter(self.sampling_params.stop_token_ids))
-            if self.eos_token_ids:
-                self.output_ids[-1] = next(iter(self.eos_token_ids))
-            self.finished_reason = FINISH_MATCHED_STR(matched="NaN happened")
-            return
+        return False
 
+    def _check_str_based_finish(self):
         if (
             len(self.sampling_params.stop_strs) > 0
             or len(self.sampling_params.stop_regex_strs) > 0
@@ -820,7 +807,7 @@ class Req:
                 for stop_str in self.sampling_params.stop_strs:
                     if stop_str in tail_str or stop_str in self.decoded_text:
                         self.finished_reason = FINISH_MATCHED_STR(matched=stop_str)
-                        return
+                        return True
 
             # Check stop regex
             if len(self.sampling_params.stop_regex_strs) > 0:
@@ -829,6 +816,57 @@ class Req:
                         self.finished_reason = FINISHED_MATCHED_REGEX(
                             matched=stop_regex_str
                         )
+                        return True
+
+        return False
+
+    def _check_vocab_boundary_finish(self, new_accepted_tokens: List[int] = None):
+        for i, token_id in enumerate(new_accepted_tokens):
+            if token_id > self.vocab_size or token_id < 0:
+                offset = len(self.output_ids) - len(new_accepted_tokens) + i
+                if self.sampling_params.stop_token_ids:
+                    self.output_ids[offset] = next(
+                        iter(self.sampling_params.stop_token_ids)
+                    )
+                if self.eos_token_ids:
+                    self.output_ids[offset] = next(iter(self.eos_token_ids))
+                self.finished_reason = FINISH_MATCHED_STR(matched="NaN happened")
+                self.finished_len = offset + 1
+                return True
+
+        return False
+
+    def check_finished(self, new_accepted_len: int = 1):
+        if self.finished():
+            return
+
+        if self.to_abort:
+            self.finished_reason = FINISH_ABORT(
+                message=self.to_abort_message,
+            )
+            return
+
+        if len(self.output_ids) >= self.sampling_params.max_new_tokens:
+            self.finished_reason = FINISH_LENGTH(
+                length=self.sampling_params.max_new_tokens
+            )
+            self.finished_len = self.sampling_params.max_new_tokens
+            return
+
+        if self.grammar is not None:
+            if self.grammar.is_terminated():
+                self.finished_reason = FINISH_MATCHED_TOKEN(matched=self.output_ids[-1])
+                return
+
+        new_accepted_tokens = self.output_ids[-new_accepted_len:]
+
+        if self._check_token_based_finish(new_accepted_tokens):
+            return
+
+        if self._check_vocab_boundary_finish(new_accepted_tokens):
+            return
+
+        if self._check_str_based_finish():
             return
 
     def reset_for_retract(self):

python/sglang/srt/managers/scheduler_output_processor_mixin.py

@@ -286,13 +286,16 @@ class SchedulerOutputProcessorMixin:
                     self.token_to_kv_pool_allocator.free(indices_to_free)
                 continue
 
+            new_accepted_len = 1
             if batch.spec_algorithm.is_none():
                 req.output_ids.append(next_token_id)
             elif batch.is_v2_eagle:
                 # Only v2 eagle's output_ids are updated here.
                 req.output_ids.extend(next_token_id)
+                new_accepted_len = len(next_token_id)
+
+            req.check_finished(new_accepted_len)
 
-            req.check_finished()
             if req.finished():
                 if batch.is_v2_eagle and self.cur_batch.forward_mode.is_extend():
                     # FIXME(lsyin): fix the messy logic here
@@ -734,6 +737,8 @@ class SchedulerOutputProcessorMixin:
                     # because of the one additional delayed token. This "continue" prevented the dummy output.
                     continue
                 req.finished_output = True
+                if req.finished_len is None:
+                    req.finished_len = len(req.output_ids)
                 should_output = True
             else:
                 if req.stream:
@@ -776,17 +781,20 @@ class SchedulerOutputProcessorMixin:
                 else:
                     decode_ids_list.append(decode_ids[req.send_decode_id_offset :])
 
+                # Exclude the tokens after stop condition
+                output_ids_ = req.output_ids_through_stop
+
                 req.send_decode_id_offset = len(decode_ids)
                 read_offsets.append(read_offset)
-                output_ids.append(req.output_ids[send_token_offset:])
-                req.send_token_offset = len(req.output_ids)
+                output_ids.append(output_ids_[send_token_offset:])
+                req.send_token_offset = len(output_ids_)
                 skip_special_tokens.append(req.sampling_params.skip_special_tokens)
                 spaces_between_special_tokens.append(
                     req.sampling_params.spaces_between_special_tokens
                 )
                 no_stop_trim.append(req.sampling_params.no_stop_trim)
                 prompt_tokens.append(len(req.origin_input_ids))
-                completion_tokens.append(len(req.output_ids))
+                completion_tokens.append(len(output_ids_))
                 cached_tokens.append(req.cached_tokens)
 
                 if not self.spec_algorithm.is_none():