Tiny rename for spec related fileds. (#18468)

python/sglang/srt/layers/attention/flashattention_backend.py

@@ -2136,7 +2136,7 @@ class FlashAttentionBackend(AttentionBackend):
                 )
             else:
                 default_extend = getattr(
-                    spec_info, "num_tokens_per_batch", self.speculative_num_steps + 1
+                    spec_info, "num_tokens_per_req", self.speculative_num_steps + 1
                 )
                 extend_seq_lens = torch.full(
                     (bs,), default_extend, dtype=torch.int32, device=device
@@ -2147,7 +2147,7 @@ class FlashAttentionBackend(AttentionBackend):
                 metadata.max_seq_len_q = int(max(extend_seq_lens_cpu))
             else:
                 metadata.max_seq_len_q = getattr(
-                    spec_info, "num_tokens_per_batch", self.speculative_num_steps + 1
+                    spec_info, "num_tokens_per_req", self.speculative_num_steps + 1
                 )
 
             metadata.cu_seqlens_q[1:].copy_(

python/sglang/srt/model_executor/forward_batch_info.py

@@ -799,7 +799,7 @@ class ForwardBatch(ForwardBatchDeepSeekMHAMixin):
                 setattr(self, "_original_batch_size", self.batch_size)
                 if self.spec_info is not None:
                     bs = self.batch_size = (
-                        num_tokens // self.spec_info.num_tokens_per_batch
+                        num_tokens // self.spec_info.num_tokens_per_req
                     )
                 else:
                     bs = self.batch_size = num_tokens
@@ -935,7 +935,7 @@ class ForwardBatch(ForwardBatchDeepSeekMHAMixin):
                 logits_output.next_token_logits = logits_output.next_token_logits[:bs]
                 logits_output.hidden_states = logits_output.hidden_states[:bs]
             elif self.forward_mode.is_draft_extend_v2():  # draft extend_v2
-                bs = bs * self.spec_info.num_tokens_per_batch
+                bs = bs * self.spec_info.num_tokens_per_req
                 logits_output.next_token_logits = logits_output.next_token_logits[:bs]
                 logits_output.hidden_states = logits_output.hidden_states[:bs]
             elif self.forward_mode.is_extend() or self.forward_mode.is_idle():

python/sglang/srt/speculative/eagle_info.py

@@ -69,7 +69,7 @@ class EagleVerifyInput(SpecInput, EagleVerifyInputV2Mixin):
     grammar: BaseGrammarObject = None
 
     # Shape info for padding
-    num_tokens_per_batch: int = -1
+    num_tokens_per_req: int = -1
 
     def __post_init__(self):
         super().__init__(SpecInputType.EAGLE_VERIFY)
@@ -634,8 +634,8 @@ class EagleDraftInput(SpecInput, EagleDraftInputV2Mixin):
     kv_indices: torch.Tensor = None
 
     # Shape info for padding
-    num_tokens_per_batch: int = -1
+    num_tokens_per_req: int = -1
-    num_tokens_for_logprob_per_batch: int = -1
+    num_tokens_for_logprob_per_req: int = -1
 
     # Inputs for draft extend
     # shape: (b,)
@@ -652,7 +652,7 @@ class EagleDraftInput(SpecInput, EagleDraftInputV2Mixin):
         super().__init__(SpecInputType.EAGLE_DRAFT)
 
     def get_spec_adjust_token_coefficient(self) -> Tuple[int, int]:
-        return self.num_tokens_per_batch, self.num_tokens_for_logprob_per_batch
+        return self.num_tokens_per_req, self.num_tokens_for_logprob_per_req
 
     def prepare_for_extend(self, batch: ScheduleBatch):
 

python/sglang/srt/speculative/eagle_info_v2.py

@@ -169,8 +169,8 @@ class EagleDraftInputV2Mixin:
             )
 
         # Get a forward batch
-        self.num_tokens_per_batch = topk
+        self.num_tokens_per_req = topk
-        self.num_tokens_for_logprob_per_batch = topk
+        self.num_tokens_for_logprob_per_req = topk
         batch.capture_hidden_mode = CaptureHiddenMode.LAST
         self.positions = batch.seq_lens.repeat_interleave(topk, dim=0)
         forward_batch = ForwardBatch.init_new(batch, draft_model_runner)

python/sglang/srt/speculative/eagle_worker.py

@@ -532,8 +532,8 @@ class EAGLEWorker(TpModelWorker):
         assert isinstance(spec_info, EagleDraftInput)
 
         spec_info.capture_hidden_mode = CaptureHiddenMode.LAST
-        spec_info.num_tokens_per_batch = self.topk
+        spec_info.num_tokens_per_req = self.topk
-        spec_info.num_tokens_for_logprob_per_batch = self.topk
+        spec_info.num_tokens_for_logprob_per_req = self.topk
         batch.return_hidden_states = False
 
         # Get forward batch
@@ -683,7 +683,7 @@ class EAGLEWorker(TpModelWorker):
     def verify(self, batch: ScheduleBatch, spec_info: EagleVerifyInput):
         seq_lens_pre_verify = batch.seq_lens.clone()
         spec_info.prepare_for_verify(batch, self.page_size)
-        spec_info.num_tokens_per_batch = self.speculative_num_steps + 1
+        spec_info.num_tokens_per_req = self.speculative_num_steps + 1
         batch.return_hidden_states = False
         batch.forward_mode = (
             ForwardMode.TARGET_VERIFY
@@ -867,8 +867,8 @@ class EAGLEWorker(TpModelWorker):
         batch.spec_info = EagleDraftInput(
             hidden_states=hidden_states,
             verified_id=next_token_ids,
-            num_tokens_per_batch=1,
+            num_tokens_per_req=1,
-            num_tokens_for_logprob_per_batch=1,
+            num_tokens_for_logprob_per_req=1,
         )
         batch.return_hidden_states = False
         batch.spec_info.prepare_for_extend(batch)
@@ -915,8 +915,8 @@ class EAGLEWorker(TpModelWorker):
                 capture_hidden_mode=CaptureHiddenMode.LAST,
             )
 
-        batch.spec_info.num_tokens_per_batch = self.speculative_num_steps + 1
+        batch.spec_info.num_tokens_per_req = self.speculative_num_steps + 1
-        batch.spec_info.num_tokens_for_logprob_per_batch = 1
+        batch.spec_info.num_tokens_for_logprob_per_req = 1
         batch.spec_info.prepare_extend_after_decode(
             batch,
             self.speculative_num_steps,

python/sglang/srt/speculative/eagle_worker_v2.py

@@ -483,9 +483,9 @@ class EagleDraftWorker(BaseDraftWorker):
             hidden_states=target_hidden_states,
             verified_id=next_token_ids,
             new_seq_lens=batch.seq_lens,
-            # draft mode is same with decode mode, only 1 num token per batch
+            # draft mode is same with decode mode, only 1 token per req
-            num_tokens_per_batch=1,
+            num_tokens_per_req=1,
-            num_tokens_for_logprob_per_batch=1,
+            num_tokens_for_logprob_per_req=1,
         )
 
         batch.spec_info = next_draft_input
@@ -508,8 +508,8 @@ class EagleDraftWorker(BaseDraftWorker):
         # Batch 2: Draft extend
         draft_input = EagleDraftInput(
             hidden_states=batch_result.logits_output.hidden_states,
-            num_tokens_per_batch=self.speculative_num_steps + 1,
+            num_tokens_per_req=self.speculative_num_steps + 1,
-            num_tokens_for_logprob_per_batch=self.speculative_num_steps + 1,
+            num_tokens_for_logprob_per_req=self.speculative_num_steps + 1,
         )
         select_index = (
             torch.arange(len(batch.seq_lens), device=self.device)
@@ -691,7 +691,7 @@ class EAGLEWorkerV2(BaseSpecWorker):
 
         # Parse args
         verify_input: EagleVerifyInput = batch.spec_info
-        verify_input.num_tokens_per_batch = self.speculative_num_steps + 1
+        verify_input.num_tokens_per_req = self.speculative_num_steps + 1
         bs = len(batch.seq_lens)
 
         # Batch 1: Target verify

python/sglang/srt/speculative/multi_layer_eagle_draft_extend_cuda_graph_runner.py

@@ -497,8 +497,8 @@ class MultiLayerEagleDraftExtendCudaGraphRunner:
 
         forward_batch.spec_info.hidden_states = self.hidden_states[:num_tokens]
         forward_batch.spec_info.accept_length = self.accept_length[:bs]
-        forward_batch.spec_info.num_tokens_per_batch = self.num_tokens_per_bs
+        forward_batch.spec_info.num_tokens_per_req = self.num_tokens_per_bs
-        forward_batch.spec_info.num_tokens_for_logprob_per_batch = 1
+        forward_batch.spec_info.num_tokens_for_logprob_per_req = 1
         forward_batch.spec_info.positions = self.positions[:num_tokens]
         forward_batch.spec_info.extend_seq_lens_tensor = self.extend_seq_lens[:bs]
 

python/sglang/srt/speculative/multi_layer_eagle_worker.py

@@ -357,8 +357,8 @@ class MultiLayerEagleWorker(TpModelWorker):
         assert isinstance(spec_info, EagleDraftInput)
 
         spec_info.capture_hidden_mode = CaptureHiddenMode.LAST
-        spec_info.num_tokens_per_batch = self.topk
+        spec_info.num_tokens_per_req = self.topk
-        spec_info.num_tokens_for_logprob_per_batch = self.topk
+        spec_info.num_tokens_for_logprob_per_req = self.topk
         batch.return_hidden_states = False
 
         # Get forward batch
@@ -599,8 +599,8 @@ class MultiLayerEagleWorker(TpModelWorker):
         batch.spec_info = EagleDraftInput(
             hidden_states=hidden_states,
             verified_id=next_token_ids,
-            num_tokens_per_batch=1,
+            num_tokens_per_req=1,
-            num_tokens_for_logprob_per_batch=1,
+            num_tokens_for_logprob_per_req=1,
         )
         batch.return_hidden_states = False
         batch.spec_info.prepare_for_extend(batch)
@@ -681,8 +681,8 @@ class MultiLayerEagleWorker(TpModelWorker):
                 capture_hidden_mode=CaptureHiddenMode.LAST,
             )
 
-        batch.spec_info.num_tokens_per_batch = self.speculative_num_steps + 1
+        batch.spec_info.num_tokens_per_req = self.speculative_num_steps + 1
-        batch.spec_info.num_tokens_for_logprob_per_batch = 1
+        batch.spec_info.num_tokens_for_logprob_per_req = 1
         batch.spec_info.prepare_extend_after_decode(
             batch,
             self.speculative_num_steps,

python/sglang/srt/speculative/multi_layer_eagle_worker_v2.py

@@ -352,9 +352,9 @@ class MultiLayerEagleDraftWorker(BaseDraftWorker):
             hidden_states=target_hidden_states,
             verified_id=next_token_ids,
             new_seq_lens=batch.seq_lens,
-            # draft mode is same with decode mode, only 1 num token per batch
+            # draft mode is same with decode mode, only 1 token per req
-            num_tokens_per_batch=1,
+            num_tokens_per_req=1,
-            num_tokens_for_logprob_per_batch=1,
+            num_tokens_for_logprob_per_req=1,
         )
 
         batch.spec_info = next_draft_input
@@ -411,8 +411,8 @@ class MultiLayerEagleDraftWorker(BaseDraftWorker):
         # Batch 2: Draft extend
         draft_input = EagleDraftInput(
             hidden_states=batch_result.logits_output.hidden_states,
-            num_tokens_per_batch=self.speculative_num_steps + 1,
+            num_tokens_per_req=self.speculative_num_steps + 1,
-            num_tokens_for_logprob_per_batch=1,
+            num_tokens_for_logprob_per_req=1,
         )
 
         # Prepare for draft extend in a separate stream