x-transformers 1.31.5__py3-none-any.whl → 1.31.7__py3-none-any.whl

x_transformers/x_transformers.py

@@ -580,7 +580,7 @@ class LayerNorm(Module):
 
     def forward(self, x):
         normed = self.ln(x)
-        gamma = self.gamma + self.unit_offset
+        gamma = self.gamma + float(self.unit_offset)
         return normed * gamma
 
 class AdaptiveLayerNorm(Module):
@@ -615,7 +615,8 @@ class ScaleNorm(Module):
         nn.init.constant_(self.g, 1. - float(unit_offset))
 
     def forward(self, x):
-        return F.normalize(x, dim = -1) * self.scale * (self.g + self.unit_offset)
+        gamma = self.g + float(self.unit_offset)
+        return F.normalize(x, dim = -1) * self.scale * gamma
 
 class RMSNorm(Module):
     def __init__(
@@ -631,7 +632,8 @@ class RMSNorm(Module):
         nn.init.constant_(self.g, 1. - float(unit_offset))
 
     def forward(self, x):
-        return F.normalize(x, dim = -1) * self.scale * (self.g + self.unit_offset)
+        gamma = self.g + float(self.unit_offset)
+        return F.normalize(x, dim = -1) * self.scale * gamma
 
 class AdaptiveRMSNorm(Module):
     def __init__(
@@ -1267,7 +1269,8 @@ class AttentionLayers(Module):
         scale_residual_constant = 1.,
         shift_tokens = 0,
         sandwich_norm = False,
-        softclamp_output_value: float | None = None,
+        softclamp_output = False,
+        softclamp_output_value = 50.,
         resi_dual = False,
         resi_dual_scale = 1.,
         zero_init_branch_output = False,
@@ -1484,6 +1487,7 @@ class AttentionLayers(Module):
         # optional soft clamping just before the final norm
         # used in gemma 2
 
+        self.softclamp_output = softclamp_output
         self.softclamp_output_value = softclamp_output_value
 
         # whether it has post norm
@@ -1717,7 +1721,7 @@ class AttentionLayers(Module):
         if return_hiddens:
             layer_hiddens.append(x)
 
-        if exists(self.softclamp_output_value):
+        if self.softclamp_output:
             x = softclamp(x, self.softclamp_output_value)
 
         final_norm = self.final_norm

{x_transformers-1.31.5.dist-info → x_transformers-1.31.7.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: x-transformers
-Version: 1.31.5
+Version: 1.31.7
 Summary: X-Transformers - Pytorch
 Home-page: https://github.com/lucidrains/x-transformers
 Author: Phil Wang

{x_transformers-1.31.5.dist-info → x_transformers-1.31.7.dist-info}/RECORD

@@ -4,11 +4,11 @@ x_transformers/autoregressive_wrapper.py,sha256=uX8Mb0zLsQrZECt_9UGt35g7tC05Rk3n
 x_transformers/continuous.py,sha256=WO52n9lFAXv5-SGadi2cApGF8dkouN8QSTEOuC7erj8,6180
 x_transformers/dpo.py,sha256=LjvWgCkqTl-UuehrzQ8nkX5guLr4whYwsmm7SKSwdls,3450
 x_transformers/nonautoregressive_wrapper.py,sha256=ys_p8obc7lTeeodCqvkRKxOXQ1C9T3j5Jwr-JbVgnXk,10432
-x_transformers/x_transformers.py,sha256=VL9Dm8L5jnpgyt_V6DWGtLs9MeiTn7ZMCQdcHSFLVo8,75871
+x_transformers/x_transformers.py,sha256=xvomb5imna2kCG_Kp-PQYsA6JGyiTx_1Dx5cD-YDlH4,75986
 x_transformers/xl_autoregressive_wrapper.py,sha256=DCx4n0_c1tFai4nOqaWVnqx2p9eutsZsDMiMP1ckxNU,4117
 x_transformers/xval.py,sha256=QE1ltYZTR_eGgIHPP2BrMWVWVLqMW-OpDZh87BSmQEg,8563
-x_transformers-1.31.5.dist-info/LICENSE,sha256=As9u198X-U-vph5noInuUfqsAG2zX_oXPHDmdjwlPPY,1066
-x_transformers-1.31.5.dist-info/METADATA,sha256=o7z9SmDOf9BAU9FyMRuAbX4LIH8nIACA0acjGHPMz8s,661
-x_transformers-1.31.5.dist-info/WHEEL,sha256=mguMlWGMX-VHnMpKOjjQidIo1ssRlCFu4a4mBpz1s2M,91
-x_transformers-1.31.5.dist-info/top_level.txt,sha256=hO6KGpFuGucRNEtRfme4A_rGcM53AKwGP7RVlRIxS5Q,15
-x_transformers-1.31.5.dist-info/RECORD,,
+x_transformers-1.31.7.dist-info/LICENSE,sha256=As9u198X-U-vph5noInuUfqsAG2zX_oXPHDmdjwlPPY,1066
+x_transformers-1.31.7.dist-info/METADATA,sha256=Z3FjZ-v02tRiDNEYG5Bqw6Yg_gjzNRBKeCWxGrmib84,661
+x_transformers-1.31.7.dist-info/WHEEL,sha256=y4mX-SOX4fYIkonsAGA5N0Oy-8_gI4FXw5HNI1xqvWg,91
+x_transformers-1.31.7.dist-info/top_level.txt,sha256=hO6KGpFuGucRNEtRfme4A_rGcM53AKwGP7RVlRIxS5Q,15
+x_transformers-1.31.7.dist-info/RECORD,,

{x_transformers-1.31.5.dist-info → x_transformers-1.31.7.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (70.1.1)
+Generator: setuptools (70.2.0)
 Root-Is-Purelib: true
 Tag: py3-none-any