Fix sliding window attention and gemma-2 unit tests in CI (#1746)

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

@@ -342,23 +342,25 @@ class FlashInferIndicesUpdaterDecode:
         for wrapper_id in range(2):
             if wrapper_id == 0:
                 # Sliding window attention
-                paged_kernel_lens = torch.minimum(  # TODO: replace this with clamp
+                paged_kernel_lens_tmp = torch.minimum(  # TODO: replace this with clamp
                     seq_lens,
                     torch.tensor(self.sliding_window_size + 1),
                 )
+                paged_kernel_lens_sum_tmp = paged_kernel_lens_tmp.sum().item()
+                kv_start_idx_tmp = seq_lens - paged_kernel_lens_tmp
             else:
                 # Full attention
-                paged_kernel_lens = seq_lens
-
-            kv_start_idx = seq_lens - paged_kernel_lens
+                paged_kernel_lens_tmp = seq_lens
+                paged_kernel_lens_sum_tmp = seq_lens_sum
+                kv_start_idx_tmp = None
 
             self.call_begin_forward(
                 decode_wrappers[wrapper_id],
                 req_pool_indices,
-                paged_kernel_lens,
-                seq_lens_sum,
+                paged_kernel_lens_tmp,
+                paged_kernel_lens_sum_tmp,
                 self.kv_indptr[wrapper_id],
-                kv_start_idx,
+                kv_start_idx_tmp,
             )
 
     def update_cross_attention(self):
@@ -369,14 +371,16 @@ class FlashInferIndicesUpdaterDecode:
         wrapper,
         req_pool_indices,
         paged_kernel_lens,
-        seq_lens_sum,
+        paged_kernel_lens_sum,
         kv_indptr,
         kv_start_idx,
     ):
         bs = len(req_pool_indices)
         kv_indptr = kv_indptr[: bs + 1]
         kv_indptr[1:] = torch.cumsum(paged_kernel_lens, dim=0)
-        kv_indices = torch.empty(seq_lens_sum, dtype=torch.int32, device="cuda")
+        kv_indices = torch.empty(
+            paged_kernel_lens_sum, dtype=torch.int32, device="cuda"
+        )
 
         create_flashinfer_kv_indices_triton[(bs,)](
             self.req_to_token,

python/sglang/test/runners.py

@@ -102,8 +102,10 @@ class HFRunner:
         return False
 
     def start_model_process(self, in_queue, out_queue, model_path, torch_dtype):
-        self.tokenizer = get_tokenizer(model_path, torch_dtype=torch.dtype)
+        # Apply model-specific patches
+        monkey_patch_gemma2_sdpa()
 
+        # Load the model and tokenizer
         if self.model_type == "generation":
             self.base_model = AutoModelForCausalLM.from_pretrained(
                 model_path,
@@ -128,7 +130,9 @@ class HFRunner:
             ).cuda()
         else:
             raise Exception(f"Unrecognized model type {self.model_type}")
+        self.tokenizer = get_tokenizer(model_path, torch_dtype=torch.dtype)
 
+        # Run forward
         while True:
             prompts, max_new_tokens, lora_paths = in_queue.get()
             if lora_paths is not None:
@@ -370,3 +374,18 @@ class SRTRunner:
     def __exit__(self, exc_type, exc_value, traceback):
         self.runtime.shutdown()
         del self.runtime
+
+
+def monkey_patch_gemma2_sdpa():
+    """
+    Use sdpa by default to fix the OOM issue.
+    Revert this commit:
+    https://github.com/huggingface/transformers/commit/975b988bfe6e7ebb47390cd9a1556c6888804883#diff-5f76eac6f18f4b491521314c318a9692318feb4d19228e9576cce7bde4240834R660
+    """
+    from transformers.models.gemma2.modeling_gemma2 import Gemma2PreTrainedModel
+
+    def _check_and_enable_sdpa(config, hard_check_only: bool = False):
+        config._attn_implementation = "sdpa"
+        return config
+
+    setattr(Gemma2PreTrainedModel, "_check_and_enable_sdpa", _check_and_enable_sdpa)