Generate: shorter XLA contrastive search tests (#20354)

This commit is contained in:
Joao Gante
2022-11-22 11:47:12 +00:00
committed by GitHub
parent c3eb01013b
commit f2e7d270ec

View File

@@ -1854,7 +1854,7 @@ class TFModelTesterMixin:
Either the model supports XLA generation and passes the inner test, or it raises an appropriate exception
"""
self._test_xla_generate(num_beams=1, num_return_sequences=1, max_new_tokens=64, penalty_alpha=0.5, top_k=4)
self._test_xla_generate(num_beams=1, num_return_sequences=1, max_new_tokens=16, penalty_alpha=0.5, top_k=4)
@slow
def test_xla_generate_slow(self):