Clean up server_args, triton cache manager (#8332)

2025-07-25 14:14:51 -07:00
parent f8260f2539
commit ed2e313eb6
12 changed files with 128 additions and 204 deletions
--- a/python/sglang/srt/speculative/eagle_draft_cuda_graph_runner.py
+++ b/python/sglang/srt/speculative/eagle_draft_cuda_graph_runner.py
@@ -336,7 +336,6 @@ class EAGLEDraftCudaGraphRunner:
            forward_batch.req_pool_indices = self.req_pool_indices[:bs]
            forward_batch.positions = self.positions[:num_tokens]

-        # Special handle for seq_len_cpu used when flashinfer mla is used
        if forward_batch.seq_lens_cpu is not None:
            if bs != raw_bs:
                self.seq_lens_cpu.fill_(self.seq_len_fill_value)