SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Fix DynamicCache and simplify Cache classes a bit (#39590)

Browse Source 
* fix

* use kwargs

* simplify

* Update cache_utils.py

* Update cache_utils.py

* Update test_cache_utils.py

* fix

* style
This commit is contained in:
Cyril Vallez
2025-07-23 10:13:45 +02:00
committed by GitHub
parent d9b35c635e
commit 5dba4bc7b2
6 changed files with 147 additions and 184 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
tests/utils/test_cache_utils.py
View File

@@ -1307,7 +1307,7 @@ class SyntheticCacheTest(unittest.TestCase):
config = copy.deepcopy(self.config)
config.num_hidden_layers = 2
config.layer_types = ["full_attention", "sliding_attention"]
config.layer_types = ["full_attention", "chunked_attention"]
config.sliding_window = 2
max_cache_len = 4
chunked_cache = HybridChunkedCache(config=config, max_batch_size=1, max_cache_len=max_cache_len)
@@ -1387,7 +1387,7 @@ class SyntheticCacheTest(unittest.TestCase):
config = copy.deepcopy(self.config)
config.num_hidden_layers = 1
config.layer_types = ["sliding_attention"]
config.layer_types = ["chunked_attention"]
config.sliding_window = 3
cache = HybridChunkedCache(config, max_batch_size=1, max_cache_len=3)