Fix ExaoneMoeMTP test that never ran in Transformers v4 (#36792)

Signed-off-by: Harry Mellor <19981378+hmellor@users.noreply.github.com>
This commit is contained in:
Harry Mellor
2026-03-11 17:10:23 +00:00
committed by GitHub
parent 196802dfa6
commit 5efa206a8c
4 changed files with 17 additions and 0 deletions
+7
View File
@@ -72,6 +72,12 @@ class _HfExamplesInfo:
If False, we will use CUDA graph and eager execution in hybrid.
"""
enable_prefix_caching: bool = True
"""
Whether to enable prefix caching for the model. If True, we will test the model with
prefix caching enabled. If False, we will test the model without prefix caching.
"""
is_available_online: bool = True
"""
Set this to `False` if the name of this architecture no longer exists on
@@ -1206,6 +1212,7 @@ _SPECULATIVE_DECODING_EXAMPLE_MODELS = {
"LGAI-EXAONE/K-EXAONE-236B-A23B",
speculative_model="LGAI-EXAONE/K-EXAONE-236B-A23B",
min_transformers_version="5.1.0",
enable_prefix_caching=False,
),
"ExtractHiddenStatesModel": _HfExamplesInfo(
"Qwen/Qwen3-8B",