support for latest transformers release 4.48.1 (#2256)

This commit is contained in:
Wing Lian
2025-01-23 21:17:57 -05:00
committed by GitHub
parent 8fb72cbc0b
commit 8a7a0b07dc
13 changed files with 98 additions and 363 deletions

View File

@@ -49,12 +49,7 @@ class TestModelPatches(unittest.TestCase):
)
normalize_config(cfg)
tokenizer = load_tokenizer(cfg)
model, _ = load_model(cfg, tokenizer, inference=False)
assert (
"MixtralFlashAttention2"
in model.model.layers[0].self_attn.__class__.__name__
)
load_model(cfg, tokenizer, inference=False)
@with_temp_dir
def test_mistral_multipack(self, temp_dir):