[Spec] fold can_run_cuda_graph into EagleVerifyOutput; drop dead extend-after-decode check#25566
Conversation
|
Warning You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again! |
|
/rerun-test test/registered/spec/eagle/test_eagle_infer_a.py test/registered/spec/eagle/test_eagle_infer_b.py test/registered/spec/eagle/test_eagle_dp_attention.py test/registered/spec/dflash/test_dflash.py test/registered/spec/test_spec_ngram.py test/registered/spec/test_spec_standalone.py test/registered/sessions/test_streaming_session_extra.py |
|
🚀 🚀 🚀 |
…rns single object
…_draft_input/is_verify_input as cross-algo phase guards; explain pp_rank=0
cfb750e to
71a0a54
Compare
|
🚀 🚀 🚀 |
71a0a54 to
6ff69cb
Compare
|
🚀 🚀 🚀 |
…nd-after-decode check (sgl-project#25566)
…nd-after-decode check (sgl-project#25566)
Rename
*VerifyInput.verify->.sample(Eagle V1, DFlash, FrozenKVMTP, Ngram); foldcan_run_cuda_graphintoEagleVerifyOutput(verify returns single object); drop deadcheck_forward_draft_extend_after_decode.CI States
Latest PR Test (Base): ⏳ Run #26030593810
Latest PR Test (Extra): ✅ Run #26030593466