x-transformers 1.35.0__py3-none-any.whl → 1.35.1__py3-none-any.whl

x_transformers/autoregressive_wrapper.py

@@ -220,13 +220,14 @@ class AutoregressiveWrapper(Module):
             if restrict_to_max_seq_len:
                 max_len_exceeded = out.shape[-1] > max_seq_len
 
-                assert not (cache_kv and max_len_exceeded and not self.net.can_cache_kv_outside_max_seq_len), 'the network cannot use cached key values when decoding outside the max sequence length. most likely because you are using absolute positional embeeding. you can switch to rotary embeddings to resolve this issue'
+                assert not (cache_kv and max_len_exceeded and not self.net.can_cache_kv_outside_max_seq_len), 'the network cannot use cached key values when decoding outside the max sequence length. most likely because you are using absolute positional embedding. you can switch to rotary embeddings to resolve this issue'
 
                 x = out[:, -max_seq_len:]
 
                 if exists(cache):
                     for inter in cache.attn_intermediates:
-                        inter.cached_kv = [t[..., -(max_seq_len - 1):, :] for t in inter.cached_kv]
+                        if inter.layer_type == 'a':
+                            inter.cached_kv = [t[..., -(max_seq_len - 1):, :] for t in inter.cached_kv]
 
             logits, new_cache = self.net(
                 x,

{x_transformers-1.35.0.dist-info → x_transformers-1.35.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: x-transformers
-Version: 1.35.0
+Version: 1.35.1
 Summary: X-Transformers - Pytorch
 Home-page: https://github.com/lucidrains/x-transformers
 Author: Phil Wang

{x_transformers-1.35.0.dist-info → x_transformers-1.35.1.dist-info}/RECORD

@@ -1,6 +1,6 @@
 x_transformers/__init__.py,sha256=-MkQrSc37cTVDX7AOykxunYnqVtFlQ7lb0Cse5dsGWU,793
 x_transformers/attend.py,sha256=7q996VGYHGIsc0FQnN8WNiwHn3xny3i1biRwx7yW5vg,12090
-x_transformers/autoregressive_wrapper.py,sha256=ka_iiej5lEBOcbutWQgGrFVMDilz2PFWzLhBh5_tmmg,10366
+x_transformers/autoregressive_wrapper.py,sha256=pDymmnPgWQoH7wwHKskI_gktsdQX-LysnQtIozodYGU,10422
 x_transformers/continuous.py,sha256=cIVEdhfei258__ziV7kQBrJMxCel54bExBTDrO9rfCI,6450
 x_transformers/dpo.py,sha256=LjvWgCkqTl-UuehrzQ8nkX5guLr4whYwsmm7SKSwdls,3450
 x_transformers/multi_input.py,sha256=tCh-fTJDj2ib4SMGtsa-AM8MxKzJAQSwqAXOu3HU2mg,9252
@@ -8,8 +8,8 @@ x_transformers/nonautoregressive_wrapper.py,sha256=ys_p8obc7lTeeodCqvkRKxOXQ1C9T
 x_transformers/x_transformers.py,sha256=2oQoQs7RMbFrVdMeOddy6yq1MhJxnficjORmMWBjjPo,80593
 x_transformers/xl_autoregressive_wrapper.py,sha256=DCx4n0_c1tFai4nOqaWVnqx2p9eutsZsDMiMP1ckxNU,4117
 x_transformers/xval.py,sha256=QE1ltYZTR_eGgIHPP2BrMWVWVLqMW-OpDZh87BSmQEg,8563
-x_transformers-1.35.0.dist-info/LICENSE,sha256=As9u198X-U-vph5noInuUfqsAG2zX_oXPHDmdjwlPPY,1066
-x_transformers-1.35.0.dist-info/METADATA,sha256=D32aQ96BsP6BXjikkuZUHc77sO6thZVO9cI_xFgLQF0,661
-x_transformers-1.35.0.dist-info/WHEEL,sha256=cVxcB9AmuTcXqmwrtPhNK88dr7IR_b6qagTj0UvIEbY,91
-x_transformers-1.35.0.dist-info/top_level.txt,sha256=hO6KGpFuGucRNEtRfme4A_rGcM53AKwGP7RVlRIxS5Q,15
-x_transformers-1.35.0.dist-info/RECORD,,
+x_transformers-1.35.1.dist-info/LICENSE,sha256=As9u198X-U-vph5noInuUfqsAG2zX_oXPHDmdjwlPPY,1066
+x_transformers-1.35.1.dist-info/METADATA,sha256=4tuROkk-uVjndRdJpYZzuvTUNC4OHgECzUinyrW14Ns,661
+x_transformers-1.35.1.dist-info/WHEEL,sha256=cVxcB9AmuTcXqmwrtPhNK88dr7IR_b6qagTj0UvIEbY,91
+x_transformers-1.35.1.dist-info/top_level.txt,sha256=hO6KGpFuGucRNEtRfme4A_rGcM53AKwGP7RVlRIxS5Q,15
+x_transformers-1.35.1.dist-info/RECORD,,