EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

Use public model for FA3 speculative decode testing (#5152)

Browse Source
This commit is contained in:
Yubo Wang
2025-04-08 00:08:25 -07:00
committed by GitHub
parent 804d9f2e4c
commit fd5a55cfd3
1 changed files with 2 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
test/srt/test_fa3.py
View File

@@ -7,8 +7,6 @@ import torch
from sglang.srt.utils import get_device_sm, kill_process_tree
from sglang.test.few_shot_gsm8k import run_eval as run_eval_few_shot_gsm8k
from sglang.test.test_utils import (
DEFAULT_EAGLE_DRAFT_MODEL_FOR_TEST,
DEFAULT_EAGLE_TARGET_MODEL_FOR_TEST,
DEFAULT_MLA_MODEL_NAME_FOR_TEST,
DEFAULT_MODEL_NAME_FOR_TEST,
DEFAULT_TIMEOUT_FOR_SERVER_LAUNCH,
@@ -125,7 +123,7 @@ class TestFlashAttention3MLA(BaseFlashAttentionTest):
class TestFlashAttention3SpeculativeDecode(BaseFlashAttentionTest):
"""Test FlashAttention3 with speculative decode enabled."""
model = DEFAULT_EAGLE_TARGET_MODEL_FOR_TEST
model = "meta-llama/Llama-3.1-8B-Instruct"
@classmethod
def get_server_args(cls):
@@ -137,7 +135,7 @@ class TestFlashAttention3SpeculativeDecode(BaseFlashAttentionTest):
"--speculative-algorithm",
"EAGLE3",
"--speculative-draft",
DEFAULT_EAGLE_DRAFT_MODEL_FOR_TEST,
"jamesliu1/sglang-EAGLE3-Llama-3.1-Instruct-8B",
"--speculative-num-steps",
"3",
"--speculative-eagle-topk",