adv-optm 1.2.dev11__py3-none-any.whl → 1.2.dev12__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Potentially problematic release.
This version of adv-optm might be problematic. Click here for more details.
- adv_optm/__init__.py +1 -1
- adv_optm/optim/Muon_adv.py +3 -1
- {adv_optm-1.2.dev11.dist-info → adv_optm-1.2.dev12.dist-info}/METADATA +1 -1
- {adv_optm-1.2.dev11.dist-info → adv_optm-1.2.dev12.dist-info}/RECORD +7 -7
- {adv_optm-1.2.dev11.dist-info → adv_optm-1.2.dev12.dist-info}/WHEEL +0 -0
- {adv_optm-1.2.dev11.dist-info → adv_optm-1.2.dev12.dist-info}/licenses/LICENSE +0 -0
- {adv_optm-1.2.dev11.dist-info → adv_optm-1.2.dev12.dist-info}/top_level.txt +0 -0
adv_optm/__init__.py
CHANGED
adv_optm/optim/Muon_adv.py
CHANGED
|
@@ -179,7 +179,9 @@ class Muon_adv(torch.optim.Optimizer):
|
|
|
179
179
|
|
|
180
180
|
# NorMuon state initialization
|
|
181
181
|
if group['normuon_variant']:
|
|
182
|
-
if
|
|
182
|
+
if state['factored']:
|
|
183
|
+
state['normuon_v'] = torch.zeros(d1, device=p.device, dtype=torch.float32)
|
|
184
|
+
elif len(p.shape) >= 2 or state['reshaped_1d_muon']:
|
|
183
185
|
num_rows = p.shape[0] if len(p.shape) >= 2 else state['effective_shape'][0]
|
|
184
186
|
state['normuon_v'] = torch.zeros(num_rows, device=p.device, dtype=torch.float32)
|
|
185
187
|
|
|
@@ -1,10 +1,10 @@
|
|
|
1
|
-
adv_optm/__init__.py,sha256=
|
|
1
|
+
adv_optm/__init__.py,sha256=eREtFkRqgGhb8_duC4ZROpRguYIQZiJfws1MJbrBn8c,380
|
|
2
2
|
adv_optm/optim/AdaMuon_adv.py,sha256=828WtdsaKXJqlZqFXE2yrsxY3Erxn-6N7CxV9jBXiaI,17880
|
|
3
3
|
adv_optm/optim/AdamW_adv.py,sha256=KL9SCJWZ_ckAQEApB6ofbndVYjancN-v7Us7hJLFf54,17475
|
|
4
4
|
adv_optm/optim/Adopt_adv.py,sha256=S8XI2YA7683jsW8p7igc2YcU30lsN0H18qL02Kpvj8E,21244
|
|
5
5
|
adv_optm/optim/Lion_Prodigy_adv.py,sha256=LEA3UYJpPeFnmxeniLNv1u2LKKj4ufx3Bq_MLw-nWXk,14617
|
|
6
6
|
adv_optm/optim/Lion_adv.py,sha256=aGNAplZlyXYgVllYcV_s4bK8iC4fv6EizFoWIMNLdBc,8299
|
|
7
|
-
adv_optm/optim/Muon_adv.py,sha256=
|
|
7
|
+
adv_optm/optim/Muon_adv.py,sha256=uo9CBUI_5hZuuKbEmqHqozvS1_d3rKK0NKlyuv_0XxU,19518
|
|
8
8
|
adv_optm/optim/Prodigy_adv.py,sha256=lEjbtuQbomsCX39DnTPeI8Z5YG0f2aZPXN_E7-nGgWw,26060
|
|
9
9
|
adv_optm/optim/Simplified_AdEMAMix.py,sha256=nEIA3yM11nBooKzHudB5l3x4UdFRBYRwiKVUkGmO0K8,12971
|
|
10
10
|
adv_optm/optim/__init__.py,sha256=hpUWE6CKtt_rvMdgQVb3PtjhfZAvAxTq6hp8H8rIpBo,489
|
|
@@ -16,8 +16,8 @@ adv_optm/util/Newton_Schulz.py,sha256=wJ_sKRaGVIsOofQ737my4ng494qX_pfgOqlDDmYtnC
|
|
|
16
16
|
adv_optm/util/One_Bit_Boolean.py,sha256=Wat49esdwohuN-OHOFMW8D0aOQgV9cP5Rl8z6yfmpos,1068
|
|
17
17
|
adv_optm/util/OrthoGrad.py,sha256=NzInuBQGy_Ja__M1R9XbvqVaQ0fhGbtGgFE9YON7B3I,707
|
|
18
18
|
adv_optm/util/__init__.py,sha256=CXzS703GB4gil85khZi7sgKOnbzXGBOltshIOSPqj18,435
|
|
19
|
-
adv_optm-1.2.
|
|
20
|
-
adv_optm-1.2.
|
|
21
|
-
adv_optm-1.2.
|
|
22
|
-
adv_optm-1.2.
|
|
23
|
-
adv_optm-1.2.
|
|
19
|
+
adv_optm-1.2.dev12.dist-info/licenses/LICENSE,sha256=HrhfyXIkWY2tGFK11kg7vPCqhgh5DcxleloqdhrpyMY,11558
|
|
20
|
+
adv_optm-1.2.dev12.dist-info/METADATA,sha256=RC88vvcjd7LgqF6wuHsBLQyoyDwm87ECoTkat7NOs5Y,14023
|
|
21
|
+
adv_optm-1.2.dev12.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
|
|
22
|
+
adv_optm-1.2.dev12.dist-info/top_level.txt,sha256=iNfBIIzu-lPrQ7jyC56WBCcbkRwitM2nJ15-MRQ_6fg,9
|
|
23
|
+
adv_optm-1.2.dev12.dist-info/RECORD,,
|
|
File without changes
|
|
File without changes
|
|
File without changes
|