adv-optm 1.2.dev15__py3-none-any.whl → 1.2.dev17__py3-none-any.whl

adv_optm/__init__.py

@@ -20,4 +20,4 @@ __all__ = [
     "AdaMuon_adv",
 ]
 
-__version__ = "1.2.dev15"
+__version__ = "1.2.dev17"

adv_optm/optim/Muon_adv.py

@@ -76,6 +76,7 @@ class Muon_adv(torch.optim.Optimizer):
         adam_beta3_ema (float): Beta3 for AdEMAMix.
         adam_alpha (float): Alpha for AdEMAMix.
         adam_kourkoutas_beta (bool): Kourkoutas-β for AdamW.
+        adam_nnmf_factor (bool): 1-bit factored for AdamW.
     """
 
     def __init__(
@@ -125,6 +126,7 @@ class Muon_adv(torch.optim.Optimizer):
         adam_ema_alpha: float = 0.95,
         adam_tiny_spike: float = 1e-9,
         adam_k_warmup_steps: int = 0,
+        adam_nnmf_factor: bool = False,
     ):
         if not (lr >= 0.0):
             raise ValueError(f"Learning-rate should be >= 0.0. Got {lr}")
@@ -165,6 +167,7 @@ class Muon_adv(torch.optim.Optimizer):
             "adam_kourkoutas_beta": adam_kourkoutas_beta, "adam_beta2_min": adam_beta2_min,
             "adam_ema_alpha": adam_ema_alpha, "adam_tiny_spike": adam_tiny_spike,
             "adam_k_warmup_steps": adam_k_warmup_steps,
+            "adam_nnmf_factor":adam_nnmf_factor,
         }
         self.stochastic_rounding = stochastic_rounding
         self.compiled_optimizer = compiled_optimizer

adv_optm/util/Kourkoutas.py

@@ -155,4 +155,5 @@ class KourkoutasHelper:
         """
         layer_key = self.optimizer.layer_key_fn(p)
         # The default is the max value, which is correct for unmapped params or edge cases
-        return self.layer_state.get(layer_key, {}).get('dynamic_beta2', group['betas'][1])
+        beta2_default = group.get('betas', group.get('adam_betas'))[1] if group.get('betas', group.get('adam_betas')) else 0.999
+        return self.layer_state.get(layer_key, {}).get('dynamic_beta2', beta2_default)

{adv_optm-1.2.dev15.dist-info → adv_optm-1.2.dev17.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: adv_optm
-Version: 1.2.dev15
+Version: 1.2.dev17
 Summary: A family of highly efficient, lightweight yet powerful optimizers.
 Home-page: https://github.com/Koratahiu/Advanced_Optimizers
 Author: Koratahiu

{adv_optm-1.2.dev15.dist-info → adv_optm-1.2.dev17.dist-info}/RECORD

@@ -1,23 +1,23 @@
-adv_optm/__init__.py,sha256=I9iRXHonvg_82dEmyKXqt9PyN04Ez8TVbMb1uZgRZAc,380
+adv_optm/__init__.py,sha256=5Dww3w78iQNwyVH82E_dmD-s6luvQjiqYS0BxKQHYCE,380
 adv_optm/optim/AdaMuon_adv.py,sha256=zjZHFS7ng5KwemQzePjFiGtNZlcgbzmmnqF6A80h_Tg,34652
 adv_optm/optim/AdamW_adv.py,sha256=KL9SCJWZ_ckAQEApB6ofbndVYjancN-v7Us7hJLFf54,17475
 adv_optm/optim/Adopt_adv.py,sha256=S8XI2YA7683jsW8p7igc2YcU30lsN0H18qL02Kpvj8E,21244
 adv_optm/optim/Lion_Prodigy_adv.py,sha256=LEA3UYJpPeFnmxeniLNv1u2LKKj4ufx3Bq_MLw-nWXk,14617
 adv_optm/optim/Lion_adv.py,sha256=aGNAplZlyXYgVllYcV_s4bK8iC4fv6EizFoWIMNLdBc,8299
-adv_optm/optim/Muon_adv.py,sha256=QutgiRkDS36O5BQNdcwdIcYBCKPy7U07YYVQT6dq8tc,33165
+adv_optm/optim/Muon_adv.py,sha256=d91wvmKKt_3IPqsqK1ZZ5cY71kuXyzy04IU3krn2NQ8,33316
 adv_optm/optim/Prodigy_adv.py,sha256=lEjbtuQbomsCX39DnTPeI8Z5YG0f2aZPXN_E7-nGgWw,26060
 adv_optm/optim/Simplified_AdEMAMix.py,sha256=nEIA3yM11nBooKzHudB5l3x4UdFRBYRwiKVUkGmO0K8,12971
 adv_optm/optim/__init__.py,sha256=hpUWE6CKtt_rvMdgQVb3PtjhfZAvAxTq6hp8H8rIpBo,489
 adv_optm/util/BF16_Stochastic_Rounding.py,sha256=Q5H0BcogmE4atP65dLoI21HKSf50lRdsBDfeF6v9Tbg,1548
 adv_optm/util/Effective_Shape.py,sha256=TBvIk1V8IuTbbBsxuekJA4e_v8JlR5Nujtut8RTWAm4,318
-adv_optm/util/Kourkoutas.py,sha256=SSzhe0B6Zb2AXGwCKpVTLr0aaFfspcFBNZCZG3azI9k,7516
+adv_optm/util/Kourkoutas.py,sha256=BnBj4WlTOJXOW0dv_vBBE27HxDTbI_1qDIWW2J7Bxdo,7644
 adv_optm/util/NNMF.py,sha256=yRf5IP5Sjq0Uf0DxN0Q8NxEGSdD-f1ULziLVDOjY8K4,639
 adv_optm/util/Newton_Schulz.py,sha256=bBboYw_jm5_FMf0Citl79uqNedkHOTjQnUI7rZgLBmY,3341
 adv_optm/util/One_Bit_Boolean.py,sha256=Wat49esdwohuN-OHOFMW8D0aOQgV9cP5Rl8z6yfmpos,1068
 adv_optm/util/OrthoGrad.py,sha256=NzInuBQGy_Ja__M1R9XbvqVaQ0fhGbtGgFE9YON7B3I,707
 adv_optm/util/__init__.py,sha256=CXzS703GB4gil85khZi7sgKOnbzXGBOltshIOSPqj18,435
-adv_optm-1.2.dev15.dist-info/licenses/LICENSE,sha256=HrhfyXIkWY2tGFK11kg7vPCqhgh5DcxleloqdhrpyMY,11558
-adv_optm-1.2.dev15.dist-info/METADATA,sha256=CH8IxEUd-TSH5dVzXKR-rl54pTIIU_JTN_MkviBWprs,14023
-adv_optm-1.2.dev15.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-adv_optm-1.2.dev15.dist-info/top_level.txt,sha256=iNfBIIzu-lPrQ7jyC56WBCcbkRwitM2nJ15-MRQ_6fg,9
-adv_optm-1.2.dev15.dist-info/RECORD,,
+adv_optm-1.2.dev17.dist-info/licenses/LICENSE,sha256=HrhfyXIkWY2tGFK11kg7vPCqhgh5DcxleloqdhrpyMY,11558
+adv_optm-1.2.dev17.dist-info/METADATA,sha256=xE_ECrY_ALerNQRFBtKml1w_n8wSp8zOH0tIz-BLiqY,14023
+adv_optm-1.2.dev17.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+adv_optm-1.2.dev17.dist-info/top_level.txt,sha256=iNfBIIzu-lPrQ7jyC56WBCcbkRwitM2nJ15-MRQ_6fg,9
+adv_optm-1.2.dev17.dist-info/RECORD,,