diff --git a/late/adamw-lr1e-5/checkpoint-1563/1_Dense/model.safetensors b/late/adamw-lr1e-5/checkpoint-1563/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..7f5fef777f9e51f190282422bb6884c64b6197b0 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-1563/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2dca8b1bc111b21a7d1ce9fe143aa12a4b4ffe77eb6bf2ac2d3fff1273c97b15 +size 9437360 diff --git a/late/adamw-lr1e-5/checkpoint-1563/2_Dense/model.safetensors b/late/adamw-lr1e-5/checkpoint-1563/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ed9fde2d65ac8358f8fccc7860f5eec9258423e7 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-1563/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:db9fae26e09e0c531089eb71cb9ac61c5e0e685d40321fb7b47c2eeb33d10619 +size 9437360 diff --git a/late/adamw-lr1e-5/checkpoint-1563/model.safetensors b/late/adamw-lr1e-5/checkpoint-1563/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ea2c296d5622340c680d30328e1465711764480c --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-1563/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ffacfe054427b76f0848f74221ee27f82b0e579e2453899b29943d6fcd6f44f5 +size 596076280 diff --git a/late/adamw-lr1e-5/checkpoint-1563/rng_state_0.pth b/late/adamw-lr1e-5/checkpoint-1563/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..26082348a981197df0510cae3337c958e26fcb6b --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-1563/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:659b1cdee2219458dd84ce6a632a595465680b8080e5c44bd600ff97eca8d752 +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-1563/rng_state_1.pth b/late/adamw-lr1e-5/checkpoint-1563/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..d46ce0452945f0c32849661b844e8c0bb9a24957 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-1563/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86accf27064cdd503053e90476a6bd10de333d4ff0594535ad55ea13a473c91d +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-1563/rng_state_2.pth b/late/adamw-lr1e-5/checkpoint-1563/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..558429d43da9b21fe074ba126896adb41ea605cb --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-1563/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18ca8d714ef40be035404c1957b5a4dee84e1f43980408393f8aa710552ee6f6 +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-1563/rng_state_3.pth b/late/adamw-lr1e-5/checkpoint-1563/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..4e54cbec0a57bff8415c48ea238e3281454ea9b4 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-1563/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2cfdebe99e40accc9c9d8f09c63136a14abda997d9b501969ec8e16e9d183179 +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-1563/scheduler.pt b/late/adamw-lr1e-5/checkpoint-1563/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..de2485473487d415a186cd8d911d0c97c4724ebf --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-1563/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2af413698366f79298aabb4c94c7e3fc93d30abe0d0ec47db82792d3b9401202 +size 1465 diff --git a/late/adamw-lr1e-5/checkpoint-1563/training_args.bin b/late/adamw-lr1e-5/checkpoint-1563/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..b62e62ab894a010eada2ef801e8f5db2e0dd5354 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-1563/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50a45d88c064a8db979a3b9de116a2f148b8ebde1374dcd07cc27cae1e683786 +size 5585 diff --git a/late/adamw-lr1e-5/checkpoint-2345/1_Dense/model.safetensors b/late/adamw-lr1e-5/checkpoint-2345/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..507a45fe2540e0cddb94716bbf17470185467b8c --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-2345/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:092640c253e1ea92f1d3437a7dc0fe961af7381441b128054b9cbc8f8d3aa713 +size 9437360 diff --git a/late/adamw-lr1e-5/checkpoint-2345/2_Dense/model.safetensors b/late/adamw-lr1e-5/checkpoint-2345/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..96885e09a29b49ee111702fa3b29e85910dbe554 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-2345/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3bdec6ac6da0f95e1496eed5da79481656220c0f3f515c5ecacfc88e49a33c2d +size 9437360 diff --git a/late/adamw-lr1e-5/checkpoint-2345/3_Dense/model.safetensors b/late/adamw-lr1e-5/checkpoint-2345/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..a8332dc7aa47eb7c18cf0f23623b6ca5728be698 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-2345/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:48b60006ac4eec2161b3d58b6ee792499dd80bf2c22fad3a736621f406049496 +size 393304 diff --git a/late/adamw-lr1e-5/checkpoint-2345/model.safetensors b/late/adamw-lr1e-5/checkpoint-2345/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6aa3a821ae086101c5e27d70532f4c0290f5e80a --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-2345/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a7b8cab13c2a8f704be436eda2d1426a72779cd9e4023652961ccf413b907a65 +size 596076280 diff --git a/late/adamw-lr1e-5/checkpoint-2345/rng_state_0.pth b/late/adamw-lr1e-5/checkpoint-2345/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..26082348a981197df0510cae3337c958e26fcb6b --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-2345/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:659b1cdee2219458dd84ce6a632a595465680b8080e5c44bd600ff97eca8d752 +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-2345/rng_state_1.pth b/late/adamw-lr1e-5/checkpoint-2345/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..d46ce0452945f0c32849661b844e8c0bb9a24957 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-2345/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86accf27064cdd503053e90476a6bd10de333d4ff0594535ad55ea13a473c91d +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-2345/rng_state_2.pth b/late/adamw-lr1e-5/checkpoint-2345/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..558429d43da9b21fe074ba126896adb41ea605cb --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-2345/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18ca8d714ef40be035404c1957b5a4dee84e1f43980408393f8aa710552ee6f6 +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-2345/rng_state_3.pth b/late/adamw-lr1e-5/checkpoint-2345/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..4e54cbec0a57bff8415c48ea238e3281454ea9b4 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-2345/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2cfdebe99e40accc9c9d8f09c63136a14abda997d9b501969ec8e16e9d183179 +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-2345/scheduler.pt b/late/adamw-lr1e-5/checkpoint-2345/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..7b358d4c83e54e35b250dcc015d797c692d56a1a --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-2345/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:72ac461e90cf57a28f5f63a7b3257ef090f6f8b574e2861dbfcfe40faa813548 +size 1465 diff --git a/late/adamw-lr1e-5/checkpoint-2345/training_args.bin b/late/adamw-lr1e-5/checkpoint-2345/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..b62e62ab894a010eada2ef801e8f5db2e0dd5354 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-2345/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50a45d88c064a8db979a3b9de116a2f148b8ebde1374dcd07cc27cae1e683786 +size 5585 diff --git a/late/adamw-lr1e-5/checkpoint-3126/rng_state_0.pth b/late/adamw-lr1e-5/checkpoint-3126/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..26082348a981197df0510cae3337c958e26fcb6b --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-3126/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:659b1cdee2219458dd84ce6a632a595465680b8080e5c44bd600ff97eca8d752 +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-3126/rng_state_1.pth b/late/adamw-lr1e-5/checkpoint-3126/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..d46ce0452945f0c32849661b844e8c0bb9a24957 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-3126/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86accf27064cdd503053e90476a6bd10de333d4ff0594535ad55ea13a473c91d +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-3126/rng_state_2.pth b/late/adamw-lr1e-5/checkpoint-3126/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..558429d43da9b21fe074ba126896adb41ea605cb --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-3126/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18ca8d714ef40be035404c1957b5a4dee84e1f43980408393f8aa710552ee6f6 +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-3126/rng_state_3.pth b/late/adamw-lr1e-5/checkpoint-3126/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..4e54cbec0a57bff8415c48ea238e3281454ea9b4 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-3126/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2cfdebe99e40accc9c9d8f09c63136a14abda997d9b501969ec8e16e9d183179 +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-3126/scheduler.pt b/late/adamw-lr1e-5/checkpoint-3126/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..114824f98145147ce973c5e16225f937fed93d83 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-3126/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c754b6fa10962e30720506300187a492fe4b2c7ab00bf88266a6cdcc3f409967 +size 1465 diff --git a/late/adamw-lr1e-5/checkpoint-3126/training_args.bin b/late/adamw-lr1e-5/checkpoint-3126/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..b62e62ab894a010eada2ef801e8f5db2e0dd5354 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-3126/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50a45d88c064a8db979a3b9de116a2f148b8ebde1374dcd07cc27cae1e683786 +size 5585 diff --git a/late/adamw-lr1e-5/checkpoint-3907/rng_state_0.pth b/late/adamw-lr1e-5/checkpoint-3907/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..26082348a981197df0510cae3337c958e26fcb6b --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-3907/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:659b1cdee2219458dd84ce6a632a595465680b8080e5c44bd600ff97eca8d752 +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-3907/rng_state_1.pth b/late/adamw-lr1e-5/checkpoint-3907/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..d46ce0452945f0c32849661b844e8c0bb9a24957 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-3907/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86accf27064cdd503053e90476a6bd10de333d4ff0594535ad55ea13a473c91d +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-3907/rng_state_2.pth b/late/adamw-lr1e-5/checkpoint-3907/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..558429d43da9b21fe074ba126896adb41ea605cb --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-3907/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18ca8d714ef40be035404c1957b5a4dee84e1f43980408393f8aa710552ee6f6 +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-3907/rng_state_3.pth b/late/adamw-lr1e-5/checkpoint-3907/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..4e54cbec0a57bff8415c48ea238e3281454ea9b4 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-3907/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2cfdebe99e40accc9c9d8f09c63136a14abda997d9b501969ec8e16e9d183179 +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-3907/scheduler.pt b/late/adamw-lr1e-5/checkpoint-3907/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..793c6e8977321529bdbbb80e4cf21b7f49aa9c38 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-3907/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f12a3d5696ed93ceeef2a45ae6b28d3cb646399484d4f361ef2729ec51b57538 +size 1465 diff --git a/late/adamw-lr1e-5/checkpoint-3907/training_args.bin b/late/adamw-lr1e-5/checkpoint-3907/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..b62e62ab894a010eada2ef801e8f5db2e0dd5354 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-3907/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50a45d88c064a8db979a3b9de116a2f148b8ebde1374dcd07cc27cae1e683786 +size 5585 diff --git a/late/adamw-lr1e-5/checkpoint-782/1_Dense/model.safetensors b/late/adamw-lr1e-5/checkpoint-782/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..966b709f5c9cf4ca1b9b6f594e2991ed54296f56 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-782/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:92a9afec4f2642afd4ef901f0713f30b786819726de4cfc1b59fc1367ea3b4bd +size 9437360 diff --git a/late/adamw-lr1e-5/checkpoint-782/2_Dense/model.safetensors b/late/adamw-lr1e-5/checkpoint-782/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c061ef2db78d0ca84d5786095e09e241cb1b8156 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-782/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0116b8d1795986be6b86c55e72a5932627c9ec46017f10c3f012538941605d73 +size 9437360 diff --git a/late/adamw-lr1e-5/checkpoint-782/3_Dense/model.safetensors b/late/adamw-lr1e-5/checkpoint-782/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b6705cb76d973c46c3f372bf3f1331915f312397 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-782/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f3d31eb333f84f524181b0acf07d05b8b470ebda6ed3fe11d3ba20af67c29ccf +size 393304 diff --git a/late/adamw-lr1e-5/checkpoint-782/model.safetensors b/late/adamw-lr1e-5/checkpoint-782/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9dd42dbaf0754a1520ae712b264b35dd0f7a7c26 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-782/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cb7927097f3952772adff2a2b6d9bc722b04c64c3e08fed1c509aaedeb12ce49 +size 596076280 diff --git a/late/adamw-lr1e-5/checkpoint-782/rng_state_0.pth b/late/adamw-lr1e-5/checkpoint-782/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..26082348a981197df0510cae3337c958e26fcb6b --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-782/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:659b1cdee2219458dd84ce6a632a595465680b8080e5c44bd600ff97eca8d752 +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-782/rng_state_1.pth b/late/adamw-lr1e-5/checkpoint-782/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..d46ce0452945f0c32849661b844e8c0bb9a24957 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-782/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86accf27064cdd503053e90476a6bd10de333d4ff0594535ad55ea13a473c91d +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-782/rng_state_2.pth b/late/adamw-lr1e-5/checkpoint-782/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..558429d43da9b21fe074ba126896adb41ea605cb --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-782/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18ca8d714ef40be035404c1957b5a4dee84e1f43980408393f8aa710552ee6f6 +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-782/rng_state_3.pth b/late/adamw-lr1e-5/checkpoint-782/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..4e54cbec0a57bff8415c48ea238e3281454ea9b4 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-782/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2cfdebe99e40accc9c9d8f09c63136a14abda997d9b501969ec8e16e9d183179 +size 15429 diff --git a/late/adamw-lr1e-5/checkpoint-782/scheduler.pt b/late/adamw-lr1e-5/checkpoint-782/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..79f33c2b8c04402be28176978a290301b239e329 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-782/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ea37e26b25c2e44b7d75d5778b0342df8216a633d012a0bf6d6065adba28ec57 +size 1465 diff --git a/late/adamw-lr1e-5/checkpoint-782/training_args.bin b/late/adamw-lr1e-5/checkpoint-782/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..b62e62ab894a010eada2ef801e8f5db2e0dd5354 --- /dev/null +++ b/late/adamw-lr1e-5/checkpoint-782/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50a45d88c064a8db979a3b9de116a2f148b8ebde1374dcd07cc27cae1e683786 +size 5585 diff --git a/late/adamw-lr1e-5/final/training_args.bin b/late/adamw-lr1e-5/final/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..b62e62ab894a010eada2ef801e8f5db2e0dd5354 --- /dev/null +++ b/late/adamw-lr1e-5/final/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50a45d88c064a8db979a3b9de116a2f148b8ebde1374dcd07cc27cae1e683786 +size 5585 diff --git a/late/muon-lr1e-4/checkpoint-1563/1_Dense/model.safetensors b/late/muon-lr1e-4/checkpoint-1563/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ba168203579feee188469fd607d1c1c425c18779 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-1563/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c6a4d4cc09c01d997fec21a567394d6469b9bc5b03fc3036a3b922ca93174fc2 +size 9437360 diff --git a/late/muon-lr1e-4/checkpoint-1563/2_Dense/model.safetensors b/late/muon-lr1e-4/checkpoint-1563/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8293903b92011d19ea02b75f1e07f77114161888 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-1563/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d5455c06c49f5e6f4b0929a87afd0c9e48701c1bef0565491ca2dd1576ebf7f4 +size 9437360 diff --git a/late/muon-lr1e-4/checkpoint-1563/3_Dense/model.safetensors b/late/muon-lr1e-4/checkpoint-1563/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b632955c1dad681c6108ff0896a3fc39805bab3d --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-1563/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e941009fb4e51be273035ece1c7fa72749f0d98610291c8c80d4c4e6c4da388e +size 393304 diff --git a/late/muon-lr1e-4/checkpoint-1563/model.safetensors b/late/muon-lr1e-4/checkpoint-1563/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..885f316ae1ee7c83b9a22d2db481e7b0caf64ebf --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-1563/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:dcebcc154fc8b52bcd9852313c9069a0f6210758338fee403b5623e18054a385 +size 596076280 diff --git a/late/muon-lr1e-4/checkpoint-1563/optimizer.pt b/late/muon-lr1e-4/checkpoint-1563/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..93d5d1e695e0f9dc59c25b611933fcbf019622c2 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-1563/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:102b0fe29247d963b15bf85eac35e4d9c7559a73776eccb9e67297e8b274b3ac +size 789544331 diff --git a/late/muon-lr1e-4/checkpoint-1563/rng_state_0.pth b/late/muon-lr1e-4/checkpoint-1563/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..9269d2c338d41e7e53c36da59ce359217c84194e --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-1563/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f91f87bd9a92e8739289f46162ef308559319763234c64c5b77db9ec569f87f +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-1563/rng_state_1.pth b/late/muon-lr1e-4/checkpoint-1563/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..85504fcf5b95eeaceb18fd0eb6e5f135811dc600 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-1563/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7d39b8b541e3fd455fdaac992f8e531a8b6ecbd342d84bd34ce103caf8c4829 +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-1563/rng_state_2.pth b/late/muon-lr1e-4/checkpoint-1563/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..b27a42d52c6b2c4abe7662f4fdaa41c28a9cc5e8 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-1563/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ff092a382271ec0416bca1eaebe1589a5c027148f661f515a99be89a41550e9 +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-1563/rng_state_3.pth b/late/muon-lr1e-4/checkpoint-1563/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..336e2fc455a1af65baad649e25993201963256e1 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-1563/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ee31a8654359133d1c40cfb2da8b7350b44cc7eec9bf923c997643420f7061d +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-1563/scheduler.pt b/late/muon-lr1e-4/checkpoint-1563/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..88552cf15499a69d8280f0f682ce73c51d6ba6bb --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-1563/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:21c3992c1d869d641783f8c82d18420047e6df98c3f879a22a409391d8942fe5 +size 1465 diff --git a/late/muon-lr1e-4/checkpoint-1563/training_args.bin b/late/muon-lr1e-4/checkpoint-1563/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..40098c989b9c264523a188c1d1a7ea536a1d6c42 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-1563/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:08095013a8dab2f40718dfcb315d5325e1d676c91bfd24b7894660369e682fd4 +size 5585 diff --git a/late/muon-lr1e-4/checkpoint-2345/1_Dense/model.safetensors b/late/muon-lr1e-4/checkpoint-2345/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b88fa035878f725988544dad9fd45bbbc205670b --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-2345/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:274cb7d41c4959b26c6a4c5e9ebb3bd0e236140ac52f842522d1dd0c40b085d9 +size 9437360 diff --git a/late/muon-lr1e-4/checkpoint-2345/2_Dense/model.safetensors b/late/muon-lr1e-4/checkpoint-2345/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0d9f771d7a5960636ad81482f6b5f9257a759c9c --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-2345/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4ef18d4e0541fe0b62c090ec2d49d04f1347775181a8a8a9c35ba334e71fbbd8 +size 9437360 diff --git a/late/muon-lr1e-4/checkpoint-2345/3_Dense/model.safetensors b/late/muon-lr1e-4/checkpoint-2345/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e63f40e5a601a173aaee69217953f2fcd43afbed --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-2345/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ad2d37d8976d579e565be55d052ece30fcf7346b3c06daacbf2fea937a664bba +size 393304 diff --git a/late/muon-lr1e-4/checkpoint-2345/model.safetensors b/late/muon-lr1e-4/checkpoint-2345/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..eae16c3e8a2c4efe8921864bae98a0eb226740e8 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-2345/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7791a8c37f29ebc7c9b8739f82408728032eab3b26ac17cabbb1e70d99bf841a +size 596076280 diff --git a/late/muon-lr1e-4/checkpoint-2345/optimizer.pt b/late/muon-lr1e-4/checkpoint-2345/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..082491dd93b6b7e52a8ff2b78569573cb71ea3e4 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-2345/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e2397e2efb6dd7c1fb06d51dea4caac28c339066df35b754ecc239f870bbcff5 +size 789544331 diff --git a/late/muon-lr1e-4/checkpoint-2345/rng_state_0.pth b/late/muon-lr1e-4/checkpoint-2345/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..9269d2c338d41e7e53c36da59ce359217c84194e --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-2345/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f91f87bd9a92e8739289f46162ef308559319763234c64c5b77db9ec569f87f +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-2345/rng_state_1.pth b/late/muon-lr1e-4/checkpoint-2345/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..85504fcf5b95eeaceb18fd0eb6e5f135811dc600 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-2345/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7d39b8b541e3fd455fdaac992f8e531a8b6ecbd342d84bd34ce103caf8c4829 +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-2345/rng_state_2.pth b/late/muon-lr1e-4/checkpoint-2345/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..b27a42d52c6b2c4abe7662f4fdaa41c28a9cc5e8 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-2345/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ff092a382271ec0416bca1eaebe1589a5c027148f661f515a99be89a41550e9 +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-2345/rng_state_3.pth b/late/muon-lr1e-4/checkpoint-2345/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..336e2fc455a1af65baad649e25993201963256e1 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-2345/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ee31a8654359133d1c40cfb2da8b7350b44cc7eec9bf923c997643420f7061d +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-2345/scheduler.pt b/late/muon-lr1e-4/checkpoint-2345/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..098ec92344fd1a27f8fb0d1a1ee41705b2d586ec --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-2345/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e996cc89434b7b9f0892d1588892cd6d6e3f7736af087ec8b5620a3ec6518cf +size 1465 diff --git a/late/muon-lr1e-4/checkpoint-2345/training_args.bin b/late/muon-lr1e-4/checkpoint-2345/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..40098c989b9c264523a188c1d1a7ea536a1d6c42 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-2345/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:08095013a8dab2f40718dfcb315d5325e1d676c91bfd24b7894660369e682fd4 +size 5585 diff --git a/late/muon-lr1e-4/checkpoint-3126/1_Dense/model.safetensors b/late/muon-lr1e-4/checkpoint-3126/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b5d198243a33a92ee3cbee840f0e0c96bdb38855 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3126/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7984c9c8afbbcb825de6be26bcac9ea855d3c3e2c3357bca655990c617c8d085 +size 9437360 diff --git a/late/muon-lr1e-4/checkpoint-3126/2_Dense/model.safetensors b/late/muon-lr1e-4/checkpoint-3126/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fb2c78a5ed7c2f2ff3ffab724c35e4f632e0bd9b --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3126/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:966f57da5ee2ee1856a99a8fb95ce9df69b4561e2a36c32693fd868b3a45afe5 +size 9437360 diff --git a/late/muon-lr1e-4/checkpoint-3126/3_Dense/model.safetensors b/late/muon-lr1e-4/checkpoint-3126/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e7ede5c5a8ec571da577020b1b118be03b89a56a --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3126/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4cc58c3bf6684fbc8b153d173ac1c015308a8437f122a47f604cf39539a293b3 +size 393304 diff --git a/late/muon-lr1e-4/checkpoint-3126/model.safetensors b/late/muon-lr1e-4/checkpoint-3126/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e4ac69966da0273a3baa05e7cf14e20277c0a476 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3126/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:665eede7bd5a79e4f29d607d8a068ed4eb99a7aa83167cd6bd6f83d2ed5a9a75 +size 596076280 diff --git a/late/muon-lr1e-4/checkpoint-3126/optimizer.pt b/late/muon-lr1e-4/checkpoint-3126/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..542dc0b203d4cffce562a530b73677bfa5da3d33 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3126/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d9e46d1b06e09fd3c5aa3f01c40c549a920f5728c9c815261be7da137ac65eb2 +size 789544331 diff --git a/late/muon-lr1e-4/checkpoint-3126/rng_state_0.pth b/late/muon-lr1e-4/checkpoint-3126/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..9269d2c338d41e7e53c36da59ce359217c84194e --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3126/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f91f87bd9a92e8739289f46162ef308559319763234c64c5b77db9ec569f87f +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-3126/rng_state_1.pth b/late/muon-lr1e-4/checkpoint-3126/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..85504fcf5b95eeaceb18fd0eb6e5f135811dc600 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3126/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7d39b8b541e3fd455fdaac992f8e531a8b6ecbd342d84bd34ce103caf8c4829 +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-3126/rng_state_2.pth b/late/muon-lr1e-4/checkpoint-3126/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..b27a42d52c6b2c4abe7662f4fdaa41c28a9cc5e8 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3126/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ff092a382271ec0416bca1eaebe1589a5c027148f661f515a99be89a41550e9 +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-3126/rng_state_3.pth b/late/muon-lr1e-4/checkpoint-3126/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..336e2fc455a1af65baad649e25993201963256e1 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3126/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ee31a8654359133d1c40cfb2da8b7350b44cc7eec9bf923c997643420f7061d +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-3126/scheduler.pt b/late/muon-lr1e-4/checkpoint-3126/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..58f14abf1e42d36d2c5d620e56f36d43cd16cee1 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3126/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:322bc19afbc71095e4583d7b6c3d9cccf5cfa93b4444f1b4f5c3750ac16cd875 +size 1465 diff --git a/late/muon-lr1e-4/checkpoint-3126/training_args.bin b/late/muon-lr1e-4/checkpoint-3126/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..40098c989b9c264523a188c1d1a7ea536a1d6c42 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3126/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:08095013a8dab2f40718dfcb315d5325e1d676c91bfd24b7894660369e682fd4 +size 5585 diff --git a/late/muon-lr1e-4/checkpoint-3907/1_Dense/model.safetensors b/late/muon-lr1e-4/checkpoint-3907/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8b532aca88a21d25b9f25f30d853b43be563699a --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3907/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6153cc9c0fec221a1a2f62c6191a872d915bd01d67e988ec73ee1e2e6b61400d +size 9437360 diff --git a/late/muon-lr1e-4/checkpoint-3907/2_Dense/model.safetensors b/late/muon-lr1e-4/checkpoint-3907/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fdda0522bf9e7b4c82dd77c004ab0b4930e1d915 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3907/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b9169100c1aa60319cc29f74658e85d282323b7faeb1078ddff36cb8c219e0c +size 9437360 diff --git a/late/muon-lr1e-4/checkpoint-3907/3_Dense/model.safetensors b/late/muon-lr1e-4/checkpoint-3907/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fa716d4b5985f92ecb5217c679c7cdd72b145b16 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3907/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:10f18a5ebc1759016e5d2c734671362fcac6b549cc5a7d27e46d5eb1ee331c1b +size 393304 diff --git a/late/muon-lr1e-4/checkpoint-3907/model.safetensors b/late/muon-lr1e-4/checkpoint-3907/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0ed7d1b40e471245c8e20a101ddf8e6b3a8f6c0e --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3907/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:849e1924ea8c6d6173e347f93666809684bbc2ba384279a4d7d2bafa7d1c3731 +size 596076280 diff --git a/late/muon-lr1e-4/checkpoint-3907/optimizer.pt b/late/muon-lr1e-4/checkpoint-3907/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..2a7ac30f9e8596ce144acbf773758291f06a4d26 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3907/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0a66968c233036bd7720a32febcfbeda31926510a4815e6558a6a566b8d8ee9d +size 789544331 diff --git a/late/muon-lr1e-4/checkpoint-3907/rng_state_0.pth b/late/muon-lr1e-4/checkpoint-3907/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..9269d2c338d41e7e53c36da59ce359217c84194e --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3907/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f91f87bd9a92e8739289f46162ef308559319763234c64c5b77db9ec569f87f +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-3907/rng_state_1.pth b/late/muon-lr1e-4/checkpoint-3907/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..85504fcf5b95eeaceb18fd0eb6e5f135811dc600 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3907/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7d39b8b541e3fd455fdaac992f8e531a8b6ecbd342d84bd34ce103caf8c4829 +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-3907/rng_state_2.pth b/late/muon-lr1e-4/checkpoint-3907/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..b27a42d52c6b2c4abe7662f4fdaa41c28a9cc5e8 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3907/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ff092a382271ec0416bca1eaebe1589a5c027148f661f515a99be89a41550e9 +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-3907/rng_state_3.pth b/late/muon-lr1e-4/checkpoint-3907/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..336e2fc455a1af65baad649e25993201963256e1 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3907/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ee31a8654359133d1c40cfb2da8b7350b44cc7eec9bf923c997643420f7061d +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-3907/scheduler.pt b/late/muon-lr1e-4/checkpoint-3907/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..7f7e27926f068177c72d39a1ed3a77464f10cd1a --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3907/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f5f98fdeb19d4db27bd14680e4cfcfeea2d1c4ecdf806b53ee763db608c2851d +size 1465 diff --git a/late/muon-lr1e-4/checkpoint-3907/training_args.bin b/late/muon-lr1e-4/checkpoint-3907/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..40098c989b9c264523a188c1d1a7ea536a1d6c42 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-3907/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:08095013a8dab2f40718dfcb315d5325e1d676c91bfd24b7894660369e682fd4 +size 5585 diff --git a/late/muon-lr1e-4/checkpoint-782/1_Dense/model.safetensors b/late/muon-lr1e-4/checkpoint-782/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e16036ec23c6968759c6511af16a700c0c8b11d9 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-782/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c8589e25c5b615e537c0e1cb28d40692d0ccc7bad20a2fababed5cf7e67956f4 +size 9437360 diff --git a/late/muon-lr1e-4/checkpoint-782/2_Dense/model.safetensors b/late/muon-lr1e-4/checkpoint-782/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..525b86c145f6ff5c90f0623bd394f5ba2d0a4e6d --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-782/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cd5aad963b2eb8696d7af815abb0da7f47db44839cd13b8da22ee26a70210d67 +size 9437360 diff --git a/late/muon-lr1e-4/checkpoint-782/3_Dense/model.safetensors b/late/muon-lr1e-4/checkpoint-782/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..1572cc0db84f243fe4ee8f21e60ced5584a68b86 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-782/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:61be370cfd4528e53faa6e2141c7a0c75d655ea0cea0d161d3c921f412f3fd85 +size 393304 diff --git a/late/muon-lr1e-4/checkpoint-782/model.safetensors b/late/muon-lr1e-4/checkpoint-782/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..4ed4810a29217973aeafaaf5a7867cfc2180c959 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-782/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6afe5deece527a61f90e2584e9c91b3ca4fee9b73d084c7c3b84888850d9aa7 +size 596076280 diff --git a/late/muon-lr1e-4/checkpoint-782/optimizer.pt b/late/muon-lr1e-4/checkpoint-782/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..586269c1792e843dc6ff67315e04b65923c583ab --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-782/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:09ce18c2e0673892d26e70d3da45ede2ddcc79c4d809be8f4cf00854a9e46875 +size 789544331 diff --git a/late/muon-lr1e-4/checkpoint-782/rng_state_0.pth b/late/muon-lr1e-4/checkpoint-782/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..9269d2c338d41e7e53c36da59ce359217c84194e --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-782/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f91f87bd9a92e8739289f46162ef308559319763234c64c5b77db9ec569f87f +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-782/rng_state_1.pth b/late/muon-lr1e-4/checkpoint-782/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..85504fcf5b95eeaceb18fd0eb6e5f135811dc600 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-782/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7d39b8b541e3fd455fdaac992f8e531a8b6ecbd342d84bd34ce103caf8c4829 +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-782/rng_state_2.pth b/late/muon-lr1e-4/checkpoint-782/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..b27a42d52c6b2c4abe7662f4fdaa41c28a9cc5e8 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-782/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ff092a382271ec0416bca1eaebe1589a5c027148f661f515a99be89a41550e9 +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-782/rng_state_3.pth b/late/muon-lr1e-4/checkpoint-782/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..336e2fc455a1af65baad649e25993201963256e1 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-782/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ee31a8654359133d1c40cfb2da8b7350b44cc7eec9bf923c997643420f7061d +size 15429 diff --git a/late/muon-lr1e-4/checkpoint-782/scheduler.pt b/late/muon-lr1e-4/checkpoint-782/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..1f78d7b60b8f04dd14da6a32cb7267f80e62e44b --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-782/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:377b8e62397a66ce7978e296453032e0cb360843fc0c934664ed0e56fde62be2 +size 1465 diff --git a/late/muon-lr1e-4/checkpoint-782/training_args.bin b/late/muon-lr1e-4/checkpoint-782/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..40098c989b9c264523a188c1d1a7ea536a1d6c42 --- /dev/null +++ b/late/muon-lr1e-4/checkpoint-782/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:08095013a8dab2f40718dfcb315d5325e1d676c91bfd24b7894660369e682fd4 +size 5585 diff --git a/late/muon-lr1e-4/final/1_Dense/model.safetensors b/late/muon-lr1e-4/final/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8b532aca88a21d25b9f25f30d853b43be563699a --- /dev/null +++ b/late/muon-lr1e-4/final/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6153cc9c0fec221a1a2f62c6191a872d915bd01d67e988ec73ee1e2e6b61400d +size 9437360 diff --git a/late/muon-lr1e-4/final/2_Dense/model.safetensors b/late/muon-lr1e-4/final/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fdda0522bf9e7b4c82dd77c004ab0b4930e1d915 --- /dev/null +++ b/late/muon-lr1e-4/final/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3b9169100c1aa60319cc29f74658e85d282323b7faeb1078ddff36cb8c219e0c +size 9437360 diff --git a/late/muon-lr1e-4/final/3_Dense/model.safetensors b/late/muon-lr1e-4/final/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fa716d4b5985f92ecb5217c679c7cdd72b145b16 --- /dev/null +++ b/late/muon-lr1e-4/final/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:10f18a5ebc1759016e5d2c734671362fcac6b549cc5a7d27e46d5eb1ee331c1b +size 393304 diff --git a/late/muon-lr1e-4/final/model.safetensors b/late/muon-lr1e-4/final/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..0ed7d1b40e471245c8e20a101ddf8e6b3a8f6c0e --- /dev/null +++ b/late/muon-lr1e-4/final/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:849e1924ea8c6d6173e347f93666809684bbc2ba384279a4d7d2bafa7d1c3731 +size 596076280 diff --git a/late/muon-lr1e-4/final/training_args.bin b/late/muon-lr1e-4/final/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..40098c989b9c264523a188c1d1a7ea536a1d6c42 --- /dev/null +++ b/late/muon-lr1e-4/final/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:08095013a8dab2f40718dfcb315d5325e1d676c91bfd24b7894660369e682fd4 +size 5585 diff --git a/late/muon-lr3e-3/checkpoint-1563/1_Dense/model.safetensors b/late/muon-lr3e-3/checkpoint-1563/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8b6f58ac125aca59de3d8c767a793d5f662d4d71 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-1563/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:28479591abf727f7ae875582de0329637dc1d67559caf50958a53d3f18f160ee +size 9437360 diff --git a/late/muon-lr3e-3/checkpoint-1563/2_Dense/model.safetensors b/late/muon-lr3e-3/checkpoint-1563/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..812703d46ad4e4f3822ede89eddbf02dd79f36a8 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-1563/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bf72053c10195b015ba4816fb4574107956db8d9f35da5dfe9fac510d30079d5 +size 9437360 diff --git a/late/muon-lr3e-3/checkpoint-1563/3_Dense/model.safetensors b/late/muon-lr3e-3/checkpoint-1563/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..a1fb2efe1ca0938e3cb992586e0e5318071f79c4 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-1563/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a2d79e9cfbc241dba59014d029bf211ff66c8c52bbc8c013038244b7d5cabd12 +size 393304 diff --git a/late/muon-lr3e-3/checkpoint-1563/model.safetensors b/late/muon-lr3e-3/checkpoint-1563/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..48436c4a81602c77f8f1b9577b4e8beb5609f428 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-1563/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9a93f1ce0e8ac775e46ce188313315395784cf79e5797f8f53902e151245fb45 +size 596076280 diff --git a/late/muon-lr3e-3/checkpoint-1563/optimizer.pt b/late/muon-lr3e-3/checkpoint-1563/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..95df859e6a2992f3f3ee3ef3574531cc06be09f3 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-1563/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41f7d56d3d36fac781c7fc77c42afb08140ca76e1a95da3ce8a01433ed170cec +size 789544331 diff --git a/late/muon-lr3e-3/checkpoint-2345/1_Dense/model.safetensors b/late/muon-lr3e-3/checkpoint-2345/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2857056bd0eeebb02e4b4738bc11ab73851f4c10 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-2345/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:14e3dbd3fa76399f2ea3c25297ae3dcb2458c7eb1d659d74580b11255adc383d +size 9437360 diff --git a/late/muon-lr3e-3/checkpoint-2345/2_Dense/model.safetensors b/late/muon-lr3e-3/checkpoint-2345/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e3381395afeaf3eab24ac3e8c67e5e46c25c76f3 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-2345/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:09481dfb426a4e060aa5b93e4637bf2f05d31d7982675f626bb8b6a9f8c12b39 +size 9437360 diff --git a/late/muon-lr3e-3/checkpoint-2345/3_Dense/model.safetensors b/late/muon-lr3e-3/checkpoint-2345/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6d2b62d02160032b3131ddd8ef3107504743ce15 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-2345/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2a250f254a3dcfe2200a529ce664ab29b5bf7b837662a28b054cd3a49f32cede +size 393304 diff --git a/late/muon-lr3e-3/checkpoint-2345/model.safetensors b/late/muon-lr3e-3/checkpoint-2345/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..fe2bfc9aa2c77a37b90244ad288d139a0c8fd7f8 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-2345/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a1779b4a7097c4fe93f62b91790de02832e9b45535534130c603269ee69bde97 +size 596076280 diff --git a/late/muon-lr3e-3/checkpoint-2345/optimizer.pt b/late/muon-lr3e-3/checkpoint-2345/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..ee37d6884d50a74ad71a9ce79eb551a7e1815efe --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-2345/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7b1bd25bb378e4a89d5ed05b35d4e1b0261b2ab6b89c672addbcbb884d139684 +size 789544331 diff --git a/late/muon-lr3e-3/checkpoint-3126/1_Dense/model.safetensors b/late/muon-lr3e-3/checkpoint-3126/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..2425e010e7016ef5348f083dc955828db9efed9a --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-3126/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ca192fcc5641a79fa7878294e1fb2da3ee4b22d7dcd7c32065e8dc83829e6a33 +size 9437360 diff --git a/late/muon-lr3e-3/checkpoint-3126/2_Dense/model.safetensors b/late/muon-lr3e-3/checkpoint-3126/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..bba91d297fa27ca09727749344dee55ff9522bda --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-3126/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c37fd98c6431752b081a90aaa6e9e5893b3d4e13c72c6de6873c838f821f0336 +size 9437360 diff --git a/late/muon-lr3e-3/checkpoint-3126/3_Dense/model.safetensors b/late/muon-lr3e-3/checkpoint-3126/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..be2e2eefba459dee454d07a6695bb0dd40790d94 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-3126/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2636ed809ffab756157d6ad586a017f15b8b9e66f1d9dcc451c525e165e8dd92 +size 393304 diff --git a/late/muon-lr3e-3/checkpoint-3126/model.safetensors b/late/muon-lr3e-3/checkpoint-3126/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e157fd08ef4bf3da808116cc52da888db15aa4ae --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-3126/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:eafb6a04ea9469f8d43fd5fe91bb3a85b7019b1e8383cdbc50bb2732eddd7b2e +size 596076280 diff --git a/late/muon-lr3e-3/checkpoint-3126/optimizer.pt b/late/muon-lr3e-3/checkpoint-3126/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..9f38c2d00ec79acc753140e98a9bd03d1ea04e48 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-3126/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aa9dbf1c570da296a127c6c320cb38059e0c35010f3042885c8373077ba0bd0e +size 789544331 diff --git a/late/muon-lr3e-3/checkpoint-3126/rng_state_0.pth b/late/muon-lr3e-3/checkpoint-3126/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..26082348a981197df0510cae3337c958e26fcb6b --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-3126/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:659b1cdee2219458dd84ce6a632a595465680b8080e5c44bd600ff97eca8d752 +size 15429 diff --git a/late/muon-lr3e-3/checkpoint-3126/rng_state_2.pth b/late/muon-lr3e-3/checkpoint-3126/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..558429d43da9b21fe074ba126896adb41ea605cb --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-3126/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18ca8d714ef40be035404c1957b5a4dee84e1f43980408393f8aa710552ee6f6 +size 15429 diff --git a/late/muon-lr3e-3/checkpoint-3126/rng_state_3.pth b/late/muon-lr3e-3/checkpoint-3126/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..4e54cbec0a57bff8415c48ea238e3281454ea9b4 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-3126/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2cfdebe99e40accc9c9d8f09c63136a14abda997d9b501969ec8e16e9d183179 +size 15429 diff --git a/late/muon-lr3e-3/checkpoint-3126/training_args.bin b/late/muon-lr3e-3/checkpoint-3126/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..bce13fc8d22934086378561aea22141f936a9933 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-3126/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bca3d1891b6ca96ea6ace696a28cfb34e4f506be6f1c769eb1261f0438857a4c +size 5585 diff --git a/late/muon-lr3e-3/checkpoint-3907/1_Dense/model.safetensors b/late/muon-lr3e-3/checkpoint-3907/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6b665e45593617f894aa6a9f4631b42a99063456 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-3907/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac64ff47dbed9e586ba102c2f0a39fe0fe726dece7da4c92e6b31fed1d8fe0c1 +size 9437360 diff --git a/late/muon-lr3e-3/checkpoint-3907/2_Dense/model.safetensors b/late/muon-lr3e-3/checkpoint-3907/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..cb5e1feb9cf5e906ea510926a0aed81e79c0fb5e --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-3907/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:26b5f14fdf1ba5b381d3349acdccbda1707f8842f8e2a2c5c8421eba2d7d7e17 +size 9437360 diff --git a/late/muon-lr3e-3/checkpoint-3907/3_Dense/model.safetensors b/late/muon-lr3e-3/checkpoint-3907/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9a0a9d1e043c786c998a1d9634e57acd89bfb993 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-3907/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e7aad15a185a8ad3f6a8b7d5f2cff7b9266effbd09d22bc231c9ae9b65169b69 +size 393304 diff --git a/late/muon-lr3e-3/checkpoint-3907/model.safetensors b/late/muon-lr3e-3/checkpoint-3907/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..55fed52caea487db6e0c5a3e1866fcee3a123204 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-3907/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:676a0c4bdc7e277b900f3cf2b2f3ccab83067c0232a92d301b235a606f598a9d +size 596076280 diff --git a/late/muon-lr3e-3/checkpoint-3907/optimizer.pt b/late/muon-lr3e-3/checkpoint-3907/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..7990c7a50bc1ffa5eaf5fd508e94d7c9d961d326 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-3907/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4c4361e98d8de079a3d8fd0571d93a45dd885ec69538279b3699f413d568eec5 +size 789544331 diff --git a/late/muon-lr3e-3/checkpoint-782/1_Dense/model.safetensors b/late/muon-lr3e-3/checkpoint-782/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..1309615f79a05597ea19214d30da1be9ad3d2826 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-782/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:611a9678ec03c0b058e6905dd830ed9af93e368546dba92e2f0615cb083ba4ea +size 9437360 diff --git a/late/muon-lr3e-3/checkpoint-782/2_Dense/model.safetensors b/late/muon-lr3e-3/checkpoint-782/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b5f0d6021dcc49a7fd8fa662f29445fe7df53230 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-782/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3cc30d657ba56cb2bfbd745cbc6bef7a8809656466adb2b48ae744478267998b +size 9437360 diff --git a/late/muon-lr3e-3/checkpoint-782/3_Dense/model.safetensors b/late/muon-lr3e-3/checkpoint-782/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..a8f97501c9a723d69dd7b18f1ae6b4d78f82a3cf --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-782/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:cad506fdb7f46893555a60a584dc975c910c38b122754b908e3a252c7109cd53 +size 393304 diff --git a/late/muon-lr3e-3/checkpoint-782/optimizer.pt b/late/muon-lr3e-3/checkpoint-782/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..33a4223795e079997b391b84743c3e04fa34c688 --- /dev/null +++ b/late/muon-lr3e-3/checkpoint-782/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5a7503e550ba41508d85b1c4374ec896019b5d85076bc6ef14d8e79de7f5d87e +size 789544331 diff --git a/late/muon-lr3e-3/final/1_Dense/model.safetensors b/late/muon-lr3e-3/final/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6b665e45593617f894aa6a9f4631b42a99063456 --- /dev/null +++ b/late/muon-lr3e-3/final/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac64ff47dbed9e586ba102c2f0a39fe0fe726dece7da4c92e6b31fed1d8fe0c1 +size 9437360 diff --git a/late/muon-lr3e-3/final/2_Dense/model.safetensors b/late/muon-lr3e-3/final/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..cb5e1feb9cf5e906ea510926a0aed81e79c0fb5e --- /dev/null +++ b/late/muon-lr3e-3/final/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:26b5f14fdf1ba5b381d3349acdccbda1707f8842f8e2a2c5c8421eba2d7d7e17 +size 9437360 diff --git a/late/muon-lr3e-3/final/3_Dense/model.safetensors b/late/muon-lr3e-3/final/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9a0a9d1e043c786c998a1d9634e57acd89bfb993 --- /dev/null +++ b/late/muon-lr3e-3/final/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e7aad15a185a8ad3f6a8b7d5f2cff7b9266effbd09d22bc231c9ae9b65169b69 +size 393304 diff --git a/late/muon-lr3e-3/final/model.safetensors b/late/muon-lr3e-3/final/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..55fed52caea487db6e0c5a3e1866fcee3a123204 --- /dev/null +++ b/late/muon-lr3e-3/final/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:676a0c4bdc7e277b900f3cf2b2f3ccab83067c0232a92d301b235a606f598a9d +size 596076280 diff --git a/late/muon-lr3e-3/final/training_args.bin b/late/muon-lr3e-3/final/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..bce13fc8d22934086378561aea22141f936a9933 --- /dev/null +++ b/late/muon-lr3e-3/final/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bca3d1891b6ca96ea6ace696a28cfb34e4f506be6f1c769eb1261f0438857a4c +size 5585 diff --git a/late/muon-lr3e-4/checkpoint-1563/1_Dense/model.safetensors b/late/muon-lr3e-4/checkpoint-1563/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f2e7649b4ebaf328cb05525d92018194c226e48d --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-1563/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a1ba26795dd34f30cb167fd3b27eec0bc1820cffc48e4eda4554a5dd786428a5 +size 9437360 diff --git a/late/muon-lr3e-4/checkpoint-1563/2_Dense/model.safetensors b/late/muon-lr3e-4/checkpoint-1563/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..7a8e1da75e6cfbe9fd9fefd443f510f8eb120afb --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-1563/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4c2ecad6048724ca4b56a81170e9e262bc74dc3ae7c116f58603f075da751499 +size 9437360 diff --git a/late/muon-lr3e-4/checkpoint-1563/3_Dense/model.safetensors b/late/muon-lr3e-4/checkpoint-1563/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..70460f49d1100631ee960acf41075126bfac52d4 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-1563/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a8bf6376b11b0dfca409b148c0731e8106b3ae971de48849329a0245fe801a58 +size 393304 diff --git a/late/muon-lr3e-4/checkpoint-1563/model.safetensors b/late/muon-lr3e-4/checkpoint-1563/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..18077f6c38a513426eaf4ac4019f8686ad427b7a --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-1563/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:867d7e6b2d6f6c9478c0bccd42e9df07d1f79132cb138d8efa0d36874ea3b818 +size 596076280 diff --git a/late/muon-lr3e-4/checkpoint-1563/optimizer.pt b/late/muon-lr3e-4/checkpoint-1563/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..52e411ee6ccd8b823adc33980759a114b3bf7c59 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-1563/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7468ed4b02dd0a0e7b839743870b38e1e7d36421ed6f9bbe55fba608a1d887d6 +size 789544331 diff --git a/late/muon-lr3e-4/checkpoint-1563/rng_state_0.pth b/late/muon-lr3e-4/checkpoint-1563/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..9269d2c338d41e7e53c36da59ce359217c84194e --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-1563/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f91f87bd9a92e8739289f46162ef308559319763234c64c5b77db9ec569f87f +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-1563/rng_state_1.pth b/late/muon-lr3e-4/checkpoint-1563/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..85504fcf5b95eeaceb18fd0eb6e5f135811dc600 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-1563/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7d39b8b541e3fd455fdaac992f8e531a8b6ecbd342d84bd34ce103caf8c4829 +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-1563/rng_state_2.pth b/late/muon-lr3e-4/checkpoint-1563/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..b27a42d52c6b2c4abe7662f4fdaa41c28a9cc5e8 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-1563/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ff092a382271ec0416bca1eaebe1589a5c027148f661f515a99be89a41550e9 +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-1563/rng_state_3.pth b/late/muon-lr3e-4/checkpoint-1563/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..336e2fc455a1af65baad649e25993201963256e1 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-1563/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ee31a8654359133d1c40cfb2da8b7350b44cc7eec9bf923c997643420f7061d +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-1563/scheduler.pt b/late/muon-lr3e-4/checkpoint-1563/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..edb362f846a5cc49e27770704aefeabc73c1103c --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-1563/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:360cbd693bcee7ec21f3ffeff2d83fc7dcb277a607da03752e535de83b7005d7 +size 1465 diff --git a/late/muon-lr3e-4/checkpoint-1563/training_args.bin b/late/muon-lr3e-4/checkpoint-1563/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6fd16c5d3736f870a19442bfadb36e8a504a56c4 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-1563/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ac0c525fe37b710736e14744765ab156ebcff884a60e18d63131664395b0c1f +size 5585 diff --git a/late/muon-lr3e-4/checkpoint-2345/1_Dense/model.safetensors b/late/muon-lr3e-4/checkpoint-2345/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c45b1e22b4e93020e9bddbe93081727f94f63b73 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-2345/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5afb394f407141e5009e51bd65712f710e54996dfd06bbbaabc90617f6cb5f26 +size 9437360 diff --git a/late/muon-lr3e-4/checkpoint-2345/2_Dense/model.safetensors b/late/muon-lr3e-4/checkpoint-2345/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..9613838d29584f260e434cc03a928cb4d7b0c190 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-2345/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3399c3be0015ba0be534e12a2cd42573beba8e8fd40acfff446255c1c8438aa0 +size 9437360 diff --git a/late/muon-lr3e-4/checkpoint-2345/3_Dense/model.safetensors b/late/muon-lr3e-4/checkpoint-2345/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ac3214762846a90d15955cf9d1671a5b786813de --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-2345/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c73b626ab57504804d4a784deb3fee0448599018e8a4303c897804033b66e3c5 +size 393304 diff --git a/late/muon-lr3e-4/checkpoint-2345/model.safetensors b/late/muon-lr3e-4/checkpoint-2345/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..de6126b2d3d55028d501d3cea7440e9fcccfb1f1 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-2345/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5e7f28e6038fa51f1ffaf9572897fa47af759ebf1e7a55f0fd8d70d0422d99e9 +size 596076280 diff --git a/late/muon-lr3e-4/checkpoint-2345/optimizer.pt b/late/muon-lr3e-4/checkpoint-2345/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..e50bcfe68e936a2b6483bb923959408cc77e5d66 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-2345/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:035d16216653cc3021b6e3f94853579422ab851a4256d8e893b0634a43e73119 +size 789544331 diff --git a/late/muon-lr3e-4/checkpoint-2345/rng_state_0.pth b/late/muon-lr3e-4/checkpoint-2345/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..9269d2c338d41e7e53c36da59ce359217c84194e --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-2345/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f91f87bd9a92e8739289f46162ef308559319763234c64c5b77db9ec569f87f +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-2345/rng_state_1.pth b/late/muon-lr3e-4/checkpoint-2345/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..85504fcf5b95eeaceb18fd0eb6e5f135811dc600 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-2345/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7d39b8b541e3fd455fdaac992f8e531a8b6ecbd342d84bd34ce103caf8c4829 +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-2345/rng_state_2.pth b/late/muon-lr3e-4/checkpoint-2345/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..b27a42d52c6b2c4abe7662f4fdaa41c28a9cc5e8 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-2345/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ff092a382271ec0416bca1eaebe1589a5c027148f661f515a99be89a41550e9 +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-2345/rng_state_3.pth b/late/muon-lr3e-4/checkpoint-2345/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..336e2fc455a1af65baad649e25993201963256e1 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-2345/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ee31a8654359133d1c40cfb2da8b7350b44cc7eec9bf923c997643420f7061d +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-2345/scheduler.pt b/late/muon-lr3e-4/checkpoint-2345/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..b2d59c5b6744b2d7ae443d76b3daf0b8ad172054 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-2345/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f7403e54de2032332db76fc886c0508deeb5626d82bcaa5d966267be07088c41 +size 1465 diff --git a/late/muon-lr3e-4/checkpoint-2345/training_args.bin b/late/muon-lr3e-4/checkpoint-2345/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6fd16c5d3736f870a19442bfadb36e8a504a56c4 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-2345/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ac0c525fe37b710736e14744765ab156ebcff884a60e18d63131664395b0c1f +size 5585 diff --git a/late/muon-lr3e-4/checkpoint-3126/1_Dense/model.safetensors b/late/muon-lr3e-4/checkpoint-3126/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..50bdb09356a523564826487fbfdfcb659120e28a --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3126/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d66aea3216d620b6ebeafaa526e09c6942ddb4bd10a538f1555fb1e10817e66a +size 9437360 diff --git a/late/muon-lr3e-4/checkpoint-3126/2_Dense/model.safetensors b/late/muon-lr3e-4/checkpoint-3126/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..945e8b0b983ac0f5abf67c90469b3bdfed0b58ba --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3126/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:396eae71e8fe3e8cdad897f783e2292bf9920b5df1e0b0d633b3a9a3dccf1ed0 +size 9437360 diff --git a/late/muon-lr3e-4/checkpoint-3126/3_Dense/model.safetensors b/late/muon-lr3e-4/checkpoint-3126/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..86e583c16928d8462bd2b8fd854fb980c9d8d81f --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3126/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3a001787736c42ad22f9d48e7028ad9ba7d18a3eaf7f565db3455bc2c57cd7d4 +size 393304 diff --git a/late/muon-lr3e-4/checkpoint-3126/model.safetensors b/late/muon-lr3e-4/checkpoint-3126/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..531c3bbef176d0f471c4e6ccbb7a8c3c84f8cfff --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3126/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d238d36c37a590101f583f7bb96fc837095b228961aacebfecd0fecc3fbc52d4 +size 596076280 diff --git a/late/muon-lr3e-4/checkpoint-3126/optimizer.pt b/late/muon-lr3e-4/checkpoint-3126/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..bb4ab2b9fa51149d77a65e69106b1076c962c610 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3126/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0f7a6a00a7353a5d99ac916d0df87af77fcde8479ea6175be76df4e6f76b97dc +size 789544331 diff --git a/late/muon-lr3e-4/checkpoint-3126/rng_state_0.pth b/late/muon-lr3e-4/checkpoint-3126/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..9269d2c338d41e7e53c36da59ce359217c84194e --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3126/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f91f87bd9a92e8739289f46162ef308559319763234c64c5b77db9ec569f87f +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-3126/rng_state_1.pth b/late/muon-lr3e-4/checkpoint-3126/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..85504fcf5b95eeaceb18fd0eb6e5f135811dc600 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3126/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7d39b8b541e3fd455fdaac992f8e531a8b6ecbd342d84bd34ce103caf8c4829 +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-3126/rng_state_2.pth b/late/muon-lr3e-4/checkpoint-3126/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..b27a42d52c6b2c4abe7662f4fdaa41c28a9cc5e8 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3126/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ff092a382271ec0416bca1eaebe1589a5c027148f661f515a99be89a41550e9 +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-3126/rng_state_3.pth b/late/muon-lr3e-4/checkpoint-3126/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..336e2fc455a1af65baad649e25993201963256e1 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3126/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ee31a8654359133d1c40cfb2da8b7350b44cc7eec9bf923c997643420f7061d +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-3126/scheduler.pt b/late/muon-lr3e-4/checkpoint-3126/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..1908322b6480575a858e4922d9cde48a48d3659b --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3126/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6d0b0b4b9e3d68546587bf379cf1eb69f243878811055338c85c13a086d4fb3 +size 1465 diff --git a/late/muon-lr3e-4/checkpoint-3126/training_args.bin b/late/muon-lr3e-4/checkpoint-3126/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6fd16c5d3736f870a19442bfadb36e8a504a56c4 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3126/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ac0c525fe37b710736e14744765ab156ebcff884a60e18d63131664395b0c1f +size 5585 diff --git a/late/muon-lr3e-4/checkpoint-3907/1_Dense/model.safetensors b/late/muon-lr3e-4/checkpoint-3907/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..d578c2d828c1c879272c3fe97e7a63b6852e9e9a --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3907/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e1824f19175ca2da53df7b665369b5582e144741a8cc3fdda22058b02e3c1d2 +size 9437360 diff --git a/late/muon-lr3e-4/checkpoint-3907/2_Dense/model.safetensors b/late/muon-lr3e-4/checkpoint-3907/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8981e41cade56ab89c54883c9f718ca45d69393d --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3907/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d3f4591d3e6a645f63cfa30839a9620636d10a42f3259db3c35acb1b117edd90 +size 9437360 diff --git a/late/muon-lr3e-4/checkpoint-3907/3_Dense/model.safetensors b/late/muon-lr3e-4/checkpoint-3907/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b041b3489fbf1662ed2194a8116351c3137794ae --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3907/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f6b3396fcc45879b2c82fa43ab213505a24080a4e6fb2449686951368885620 +size 393304 diff --git a/late/muon-lr3e-4/checkpoint-3907/model.safetensors b/late/muon-lr3e-4/checkpoint-3907/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f1f158d46685cffdff6a78e4048b1786847c2867 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3907/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d5a4a7c9e4471f1f05db3b2b447c410de9ab49f7a32daa78bd49096f6a1de377 +size 596076280 diff --git a/late/muon-lr3e-4/checkpoint-3907/optimizer.pt b/late/muon-lr3e-4/checkpoint-3907/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..1ca132384e3a758e0903406c3c999564f450eb80 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3907/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d8d738a84969020323d346a946e857ae70f69c017cda13f6fbd07a19048f4bba +size 789544331 diff --git a/late/muon-lr3e-4/checkpoint-3907/rng_state_0.pth b/late/muon-lr3e-4/checkpoint-3907/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..9269d2c338d41e7e53c36da59ce359217c84194e --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3907/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f91f87bd9a92e8739289f46162ef308559319763234c64c5b77db9ec569f87f +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-3907/rng_state_1.pth b/late/muon-lr3e-4/checkpoint-3907/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..85504fcf5b95eeaceb18fd0eb6e5f135811dc600 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3907/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7d39b8b541e3fd455fdaac992f8e531a8b6ecbd342d84bd34ce103caf8c4829 +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-3907/rng_state_2.pth b/late/muon-lr3e-4/checkpoint-3907/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..b27a42d52c6b2c4abe7662f4fdaa41c28a9cc5e8 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3907/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ff092a382271ec0416bca1eaebe1589a5c027148f661f515a99be89a41550e9 +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-3907/rng_state_3.pth b/late/muon-lr3e-4/checkpoint-3907/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..336e2fc455a1af65baad649e25993201963256e1 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3907/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ee31a8654359133d1c40cfb2da8b7350b44cc7eec9bf923c997643420f7061d +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-3907/scheduler.pt b/late/muon-lr3e-4/checkpoint-3907/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..bcb7fdba9a12c28b33aa3b861365b7b6dab5682f --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3907/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:879a8f3cfdcd0434599862cd7a83b7cfcbcae0a304fea20d87530b306c92ce34 +size 1465 diff --git a/late/muon-lr3e-4/checkpoint-3907/training_args.bin b/late/muon-lr3e-4/checkpoint-3907/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6fd16c5d3736f870a19442bfadb36e8a504a56c4 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-3907/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ac0c525fe37b710736e14744765ab156ebcff884a60e18d63131664395b0c1f +size 5585 diff --git a/late/muon-lr3e-4/checkpoint-782/1_Dense/model.safetensors b/late/muon-lr3e-4/checkpoint-782/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..63c087aa1772b92b2893aa629b01121ee3fa77cd --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-782/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4bc9b24eb9c47cd3fe1ba299e517d9534c4d4b121913ce73e324a5c27ba5788d +size 9437360 diff --git a/late/muon-lr3e-4/checkpoint-782/2_Dense/model.safetensors b/late/muon-lr3e-4/checkpoint-782/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..721bb92fd17408842d2dae77f02860493bfb3e12 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-782/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f1af354de292edeea36d22b91bf49ad5ae8e8b470c407ed5a2bf623f8d765a15 +size 9437360 diff --git a/late/muon-lr3e-4/checkpoint-782/3_Dense/model.safetensors b/late/muon-lr3e-4/checkpoint-782/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..67173003b67629192e357b6525ea8db65845eb44 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-782/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:41ce1af8bcb8b4bf868f86e859e9c2455c8058d5462f6454221714c9ca35e3a2 +size 393304 diff --git a/late/muon-lr3e-4/checkpoint-782/model.safetensors b/late/muon-lr3e-4/checkpoint-782/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e398b1726ad520d8374ff0e3573f4b9cccfcede7 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-782/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ff35db060fb5a11035316fdb8e79ba708d985268985258f8c601b3ba1e7436c1 +size 596076280 diff --git a/late/muon-lr3e-4/checkpoint-782/optimizer.pt b/late/muon-lr3e-4/checkpoint-782/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..c86f8b3f4e2345e482b1a351501519c13ed7985e --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-782/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:597c7a5061e504cd730adff37badf3d90fd13614a8680a80b85140243fec4a3c +size 789544331 diff --git a/late/muon-lr3e-4/checkpoint-782/rng_state_0.pth b/late/muon-lr3e-4/checkpoint-782/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..9269d2c338d41e7e53c36da59ce359217c84194e --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-782/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7f91f87bd9a92e8739289f46162ef308559319763234c64c5b77db9ec569f87f +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-782/rng_state_1.pth b/late/muon-lr3e-4/checkpoint-782/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..85504fcf5b95eeaceb18fd0eb6e5f135811dc600 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-782/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d7d39b8b541e3fd455fdaac992f8e531a8b6ecbd342d84bd34ce103caf8c4829 +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-782/rng_state_2.pth b/late/muon-lr3e-4/checkpoint-782/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..b27a42d52c6b2c4abe7662f4fdaa41c28a9cc5e8 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-782/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0ff092a382271ec0416bca1eaebe1589a5c027148f661f515a99be89a41550e9 +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-782/rng_state_3.pth b/late/muon-lr3e-4/checkpoint-782/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..336e2fc455a1af65baad649e25993201963256e1 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-782/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ee31a8654359133d1c40cfb2da8b7350b44cc7eec9bf923c997643420f7061d +size 15429 diff --git a/late/muon-lr3e-4/checkpoint-782/scheduler.pt b/late/muon-lr3e-4/checkpoint-782/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..c96625b83b1aa9282c015aaf59b10fdf33e55586 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-782/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9d79afa14b518c4a1c25334b3df04eede9303826eee166a76c1efdb5f46bcc6d +size 1465 diff --git a/late/muon-lr3e-4/checkpoint-782/training_args.bin b/late/muon-lr3e-4/checkpoint-782/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6fd16c5d3736f870a19442bfadb36e8a504a56c4 --- /dev/null +++ b/late/muon-lr3e-4/checkpoint-782/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ac0c525fe37b710736e14744765ab156ebcff884a60e18d63131664395b0c1f +size 5585 diff --git a/late/muon-lr3e-4/final/1_Dense/model.safetensors b/late/muon-lr3e-4/final/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..d578c2d828c1c879272c3fe97e7a63b6852e9e9a --- /dev/null +++ b/late/muon-lr3e-4/final/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e1824f19175ca2da53df7b665369b5582e144741a8cc3fdda22058b02e3c1d2 +size 9437360 diff --git a/late/muon-lr3e-4/final/2_Dense/model.safetensors b/late/muon-lr3e-4/final/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8981e41cade56ab89c54883c9f718ca45d69393d --- /dev/null +++ b/late/muon-lr3e-4/final/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d3f4591d3e6a645f63cfa30839a9620636d10a42f3259db3c35acb1b117edd90 +size 9437360 diff --git a/late/muon-lr3e-4/final/3_Dense/model.safetensors b/late/muon-lr3e-4/final/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..b041b3489fbf1662ed2194a8116351c3137794ae --- /dev/null +++ b/late/muon-lr3e-4/final/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8f6b3396fcc45879b2c82fa43ab213505a24080a4e6fb2449686951368885620 +size 393304 diff --git a/late/muon-lr3e-4/final/model.safetensors b/late/muon-lr3e-4/final/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f1f158d46685cffdff6a78e4048b1786847c2867 --- /dev/null +++ b/late/muon-lr3e-4/final/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d5a4a7c9e4471f1f05db3b2b447c410de9ab49f7a32daa78bd49096f6a1de377 +size 596076280 diff --git a/late/muon-lr3e-4/final/training_args.bin b/late/muon-lr3e-4/final/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..6fd16c5d3736f870a19442bfadb36e8a504a56c4 --- /dev/null +++ b/late/muon-lr3e-4/final/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5ac0c525fe37b710736e14744765ab156ebcff884a60e18d63131664395b0c1f +size 5585 diff --git a/late/normuon-lr3e-4/checkpoint-1563/1_Dense/model.safetensors b/late/normuon-lr3e-4/checkpoint-1563/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..8bdb1cdc3bd7941ef7d2a87af068db2de2f188a1 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-1563/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:95531793cbd1a75781f001a8bfba9ac5ecb3c3ea91cf4b7cb555bcc5dc65427c +size 9437360 diff --git a/late/normuon-lr3e-4/checkpoint-1563/2_Dense/model.safetensors b/late/normuon-lr3e-4/checkpoint-1563/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..ba3a6fd15a44e5f8e13acb58e04e5a43feccec0d --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-1563/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e32fc80d439bec042c9f30016c0cce7ad43608186649b2b86ba50116f6c5f13d +size 9437360 diff --git a/late/normuon-lr3e-4/checkpoint-1563/3_Dense/model.safetensors b/late/normuon-lr3e-4/checkpoint-1563/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..be7efc6aac47d1759e2fdd78c0b8af7c20d816b8 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-1563/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fa93fd57b47db8c615ff3e4a22762b92cbf9700c1484c0224486ed5e540112e8 +size 393304 diff --git a/late/normuon-lr3e-4/checkpoint-1563/model.safetensors b/late/normuon-lr3e-4/checkpoint-1563/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..7b154e85bb1ab47ec02680592d532a420cb520aa --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-1563/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3eadea8a7a6d87db4a525b0e4252440c1527d5249c1d7299b72cc416b9d5e121 +size 596076280 diff --git a/late/normuon-lr3e-4/checkpoint-1563/optimizer.pt b/late/normuon-lr3e-4/checkpoint-1563/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..bca04f266a08d4b97fc0a11e5d79a870162c419c --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-1563/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:adb8bcde75355543e5d5356e090a1635fb7a191c3e1da3fc062962ccda9ef178 +size 790110283 diff --git a/late/normuon-lr3e-4/checkpoint-1563/rng_state_0.pth b/late/normuon-lr3e-4/checkpoint-1563/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..26082348a981197df0510cae3337c958e26fcb6b --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-1563/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:659b1cdee2219458dd84ce6a632a595465680b8080e5c44bd600ff97eca8d752 +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-1563/rng_state_1.pth b/late/normuon-lr3e-4/checkpoint-1563/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..d46ce0452945f0c32849661b844e8c0bb9a24957 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-1563/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86accf27064cdd503053e90476a6bd10de333d4ff0594535ad55ea13a473c91d +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-1563/rng_state_2.pth b/late/normuon-lr3e-4/checkpoint-1563/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..558429d43da9b21fe074ba126896adb41ea605cb --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-1563/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18ca8d714ef40be035404c1957b5a4dee84e1f43980408393f8aa710552ee6f6 +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-1563/rng_state_3.pth b/late/normuon-lr3e-4/checkpoint-1563/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..4e54cbec0a57bff8415c48ea238e3281454ea9b4 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-1563/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2cfdebe99e40accc9c9d8f09c63136a14abda997d9b501969ec8e16e9d183179 +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-1563/scheduler.pt b/late/normuon-lr3e-4/checkpoint-1563/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..edb362f846a5cc49e27770704aefeabc73c1103c --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-1563/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:360cbd693bcee7ec21f3ffeff2d83fc7dcb277a607da03752e535de83b7005d7 +size 1465 diff --git a/late/normuon-lr3e-4/checkpoint-1563/training_args.bin b/late/normuon-lr3e-4/checkpoint-1563/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..7bf9d67c18fa3ee38a9cb11071064acedbac8a59 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-1563/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0941fa1a7411f123f60e2e837ba5997d21a805b9d232f024ba98588cc37de87c +size 5585 diff --git a/late/normuon-lr3e-4/checkpoint-2345/1_Dense/model.safetensors b/late/normuon-lr3e-4/checkpoint-2345/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f9dde2e6b7c8ff71d073ad24bec68aae0e76c5d9 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-2345/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f68cc9701b9c14010ec87b7e77eac85d3fba18290106888d1e6e475cd4cf0f17 +size 9437360 diff --git a/late/normuon-lr3e-4/checkpoint-2345/2_Dense/model.safetensors b/late/normuon-lr3e-4/checkpoint-2345/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..3842e069aeb7d6e5fccd95c4c4d0983e9d405b0c --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-2345/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ee27792613dacf886294416269aa6830eaa943227fd6d9abcee6038e9acba77a +size 9437360 diff --git a/late/normuon-lr3e-4/checkpoint-2345/3_Dense/model.safetensors b/late/normuon-lr3e-4/checkpoint-2345/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..7912d585b3e925bc6721ea554025ab65004a9bfb --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-2345/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:734b0e00f62165dec74cbeac53fcedab9f2ca9aa1e1d5f156cbc18ff4cba1459 +size 393304 diff --git a/late/normuon-lr3e-4/checkpoint-2345/model.safetensors b/late/normuon-lr3e-4/checkpoint-2345/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..69bc5937a8192365b5b5ea426721703915907234 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-2345/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f472f0b8226dbb459367bd4af1e7ebd84184425d38cfec58000da5e563d66c28 +size 596076280 diff --git a/late/normuon-lr3e-4/checkpoint-2345/optimizer.pt b/late/normuon-lr3e-4/checkpoint-2345/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..18a993029235ce7317dc6106f95c4bd3ac2bb73d --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-2345/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d23e0b2060a0b4b88bd47b37b7f1a4bdec31c6d4a392a4bba01be06b68cf1992 +size 790110283 diff --git a/late/normuon-lr3e-4/checkpoint-2345/rng_state_0.pth b/late/normuon-lr3e-4/checkpoint-2345/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..26082348a981197df0510cae3337c958e26fcb6b --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-2345/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:659b1cdee2219458dd84ce6a632a595465680b8080e5c44bd600ff97eca8d752 +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-2345/rng_state_1.pth b/late/normuon-lr3e-4/checkpoint-2345/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..d46ce0452945f0c32849661b844e8c0bb9a24957 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-2345/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86accf27064cdd503053e90476a6bd10de333d4ff0594535ad55ea13a473c91d +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-2345/rng_state_2.pth b/late/normuon-lr3e-4/checkpoint-2345/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..558429d43da9b21fe074ba126896adb41ea605cb --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-2345/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18ca8d714ef40be035404c1957b5a4dee84e1f43980408393f8aa710552ee6f6 +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-2345/rng_state_3.pth b/late/normuon-lr3e-4/checkpoint-2345/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..4e54cbec0a57bff8415c48ea238e3281454ea9b4 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-2345/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2cfdebe99e40accc9c9d8f09c63136a14abda997d9b501969ec8e16e9d183179 +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-2345/scheduler.pt b/late/normuon-lr3e-4/checkpoint-2345/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..b2d59c5b6744b2d7ae443d76b3daf0b8ad172054 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-2345/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f7403e54de2032332db76fc886c0508deeb5626d82bcaa5d966267be07088c41 +size 1465 diff --git a/late/normuon-lr3e-4/checkpoint-2345/training_args.bin b/late/normuon-lr3e-4/checkpoint-2345/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..7bf9d67c18fa3ee38a9cb11071064acedbac8a59 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-2345/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0941fa1a7411f123f60e2e837ba5997d21a805b9d232f024ba98588cc37de87c +size 5585 diff --git a/late/normuon-lr3e-4/checkpoint-3126/1_Dense/model.safetensors b/late/normuon-lr3e-4/checkpoint-3126/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..d1d6107c1fcb30a6aeddbd8a69e5397851d62ca2 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3126/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:99fffb0c183e98b0a6fcf6c6327804a2fe2cc349be4a2b88778edc5c17a37503 +size 9437360 diff --git a/late/normuon-lr3e-4/checkpoint-3126/2_Dense/model.safetensors b/late/normuon-lr3e-4/checkpoint-3126/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..74a589168961018042e54a3ba45bc537597e306d --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3126/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:25798f1bfcdc2596146782aba010c145c54a81274bca105bf2d3cc5d9407d41c +size 9437360 diff --git a/late/normuon-lr3e-4/checkpoint-3126/3_Dense/model.safetensors b/late/normuon-lr3e-4/checkpoint-3126/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c4b463d9773cf3c992cdf64ed0d694140c9d0fb9 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3126/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ac5fa2a58bae8c66776358a2b01c32375585630c82cdf8a8574920cf7fbb86d6 +size 393304 diff --git a/late/normuon-lr3e-4/checkpoint-3126/model.safetensors b/late/normuon-lr3e-4/checkpoint-3126/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..6ae1ad4e6d6fa8169f0b60b3619d3864888a8fff --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3126/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6f290a643cb9dd76c5fe0fc58445ed8736fcc936a92bc83979bacee460d6ef7f +size 596076280 diff --git a/late/normuon-lr3e-4/checkpoint-3126/optimizer.pt b/late/normuon-lr3e-4/checkpoint-3126/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..80c734c949d81014b641e4c71cc2caff0c234317 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3126/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fc202ac026a4de08d31079dcddba36d622ce37a33b1626eedbabd4aca8cf0008 +size 790110283 diff --git a/late/normuon-lr3e-4/checkpoint-3126/rng_state_0.pth b/late/normuon-lr3e-4/checkpoint-3126/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..26082348a981197df0510cae3337c958e26fcb6b --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3126/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:659b1cdee2219458dd84ce6a632a595465680b8080e5c44bd600ff97eca8d752 +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-3126/rng_state_1.pth b/late/normuon-lr3e-4/checkpoint-3126/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..d46ce0452945f0c32849661b844e8c0bb9a24957 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3126/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86accf27064cdd503053e90476a6bd10de333d4ff0594535ad55ea13a473c91d +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-3126/rng_state_2.pth b/late/normuon-lr3e-4/checkpoint-3126/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..558429d43da9b21fe074ba126896adb41ea605cb --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3126/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18ca8d714ef40be035404c1957b5a4dee84e1f43980408393f8aa710552ee6f6 +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-3126/rng_state_3.pth b/late/normuon-lr3e-4/checkpoint-3126/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..4e54cbec0a57bff8415c48ea238e3281454ea9b4 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3126/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2cfdebe99e40accc9c9d8f09c63136a14abda997d9b501969ec8e16e9d183179 +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-3126/scheduler.pt b/late/normuon-lr3e-4/checkpoint-3126/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..1908322b6480575a858e4922d9cde48a48d3659b --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3126/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f6d0b0b4b9e3d68546587bf379cf1eb69f243878811055338c85c13a086d4fb3 +size 1465 diff --git a/late/normuon-lr3e-4/checkpoint-3126/training_args.bin b/late/normuon-lr3e-4/checkpoint-3126/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..7bf9d67c18fa3ee38a9cb11071064acedbac8a59 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3126/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0941fa1a7411f123f60e2e837ba5997d21a805b9d232f024ba98588cc37de87c +size 5585 diff --git a/late/normuon-lr3e-4/checkpoint-3907/1_Dense/model.safetensors b/late/normuon-lr3e-4/checkpoint-3907/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e7de512a4349e6a19496e32552ec486f90940710 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3907/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:59e03d84c8da56e8ae900a2b7daae1164c924b5ccbceaba7ad7c0f7ac3bce76f +size 9437360 diff --git a/late/normuon-lr3e-4/checkpoint-3907/2_Dense/model.safetensors b/late/normuon-lr3e-4/checkpoint-3907/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c26ea30eeb1d682e62c0fbd7b923f7834780aae4 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3907/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:711b449a3e768bc46f83d66c3167bdb3fb92207ffcb668ff7f0294e0e5a9e0b6 +size 9437360 diff --git a/late/normuon-lr3e-4/checkpoint-3907/3_Dense/model.safetensors b/late/normuon-lr3e-4/checkpoint-3907/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..704f5c21542a7ed7d3d014a18e68cca75f44a174 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3907/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:749087267b427a0be23668032e014b0759efef180707664940190eb5b0a02f58 +size 393304 diff --git a/late/normuon-lr3e-4/checkpoint-3907/model.safetensors b/late/normuon-lr3e-4/checkpoint-3907/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..86f07deec50c465c151b63be36773402430159fa --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3907/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:243fc9b3f2bf883e408347786a3b57e23703a817449be02e6c9b7b3592ab3656 +size 596076280 diff --git a/late/normuon-lr3e-4/checkpoint-3907/optimizer.pt b/late/normuon-lr3e-4/checkpoint-3907/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..9179c2dbb99dd3e496bce7cfa437b957ca9bb147 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3907/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:25862eb7e25f1c517f245824557dc0b9db8a881ffc01ee8ca7e8850921a4d439 +size 790110283 diff --git a/late/normuon-lr3e-4/checkpoint-3907/rng_state_0.pth b/late/normuon-lr3e-4/checkpoint-3907/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..26082348a981197df0510cae3337c958e26fcb6b --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3907/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:659b1cdee2219458dd84ce6a632a595465680b8080e5c44bd600ff97eca8d752 +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-3907/rng_state_1.pth b/late/normuon-lr3e-4/checkpoint-3907/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..d46ce0452945f0c32849661b844e8c0bb9a24957 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3907/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86accf27064cdd503053e90476a6bd10de333d4ff0594535ad55ea13a473c91d +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-3907/rng_state_2.pth b/late/normuon-lr3e-4/checkpoint-3907/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..558429d43da9b21fe074ba126896adb41ea605cb --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3907/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18ca8d714ef40be035404c1957b5a4dee84e1f43980408393f8aa710552ee6f6 +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-3907/rng_state_3.pth b/late/normuon-lr3e-4/checkpoint-3907/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..4e54cbec0a57bff8415c48ea238e3281454ea9b4 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3907/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2cfdebe99e40accc9c9d8f09c63136a14abda997d9b501969ec8e16e9d183179 +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-3907/scheduler.pt b/late/normuon-lr3e-4/checkpoint-3907/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..bcb7fdba9a12c28b33aa3b861365b7b6dab5682f --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3907/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:879a8f3cfdcd0434599862cd7a83b7cfcbcae0a304fea20d87530b306c92ce34 +size 1465 diff --git a/late/normuon-lr3e-4/checkpoint-3907/training_args.bin b/late/normuon-lr3e-4/checkpoint-3907/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..7bf9d67c18fa3ee38a9cb11071064acedbac8a59 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-3907/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0941fa1a7411f123f60e2e837ba5997d21a805b9d232f024ba98588cc37de87c +size 5585 diff --git a/late/normuon-lr3e-4/checkpoint-782/1_Dense/model.safetensors b/late/normuon-lr3e-4/checkpoint-782/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..cbc9b0a4117414f1feebd3d90420754227a014f6 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-782/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ed2f08d22948e148470fa640cba0369e99ef6e385727415182369a4b5530eb15 +size 9437360 diff --git a/late/normuon-lr3e-4/checkpoint-782/2_Dense/model.safetensors b/late/normuon-lr3e-4/checkpoint-782/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..070523befe5a8a37120490c5971bdb9972bce033 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-782/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f1467885b5d8bddfa1cc337130eef084281ccdd52a6621a4702932599ad21f89 +size 9437360 diff --git a/late/normuon-lr3e-4/checkpoint-782/3_Dense/model.safetensors b/late/normuon-lr3e-4/checkpoint-782/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..f23a26fc008babecd755bc2042d5fa7a99efa563 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-782/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f68ceb04f8b8e0e00b0ee78035222f2acae95f8c2326a97cff3665e07e014db5 +size 393304 diff --git a/late/normuon-lr3e-4/checkpoint-782/model.safetensors b/late/normuon-lr3e-4/checkpoint-782/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..4dcc0172fc1c4eb31f8c7cecb85907ce771bb266 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-782/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0df25e056028e4a77ca8f7e74415b3feba58dd4ecd588b43aebf7fc1c8abc11c +size 596076280 diff --git a/late/normuon-lr3e-4/checkpoint-782/optimizer.pt b/late/normuon-lr3e-4/checkpoint-782/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..c6ba843271061b7836ebaabced6befef8da75b6c --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-782/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6748253f74760341b650e328d1671644085af8b22af9dc57cec035b1bb205eeb +size 790110283 diff --git a/late/normuon-lr3e-4/checkpoint-782/rng_state_0.pth b/late/normuon-lr3e-4/checkpoint-782/rng_state_0.pth new file mode 100644 index 0000000000000000000000000000000000000000..26082348a981197df0510cae3337c958e26fcb6b --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-782/rng_state_0.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:659b1cdee2219458dd84ce6a632a595465680b8080e5c44bd600ff97eca8d752 +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-782/rng_state_1.pth b/late/normuon-lr3e-4/checkpoint-782/rng_state_1.pth new file mode 100644 index 0000000000000000000000000000000000000000..d46ce0452945f0c32849661b844e8c0bb9a24957 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-782/rng_state_1.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86accf27064cdd503053e90476a6bd10de333d4ff0594535ad55ea13a473c91d +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-782/rng_state_2.pth b/late/normuon-lr3e-4/checkpoint-782/rng_state_2.pth new file mode 100644 index 0000000000000000000000000000000000000000..558429d43da9b21fe074ba126896adb41ea605cb --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-782/rng_state_2.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:18ca8d714ef40be035404c1957b5a4dee84e1f43980408393f8aa710552ee6f6 +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-782/rng_state_3.pth b/late/normuon-lr3e-4/checkpoint-782/rng_state_3.pth new file mode 100644 index 0000000000000000000000000000000000000000..4e54cbec0a57bff8415c48ea238e3281454ea9b4 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-782/rng_state_3.pth @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2cfdebe99e40accc9c9d8f09c63136a14abda997d9b501969ec8e16e9d183179 +size 15429 diff --git a/late/normuon-lr3e-4/checkpoint-782/scheduler.pt b/late/normuon-lr3e-4/checkpoint-782/scheduler.pt new file mode 100644 index 0000000000000000000000000000000000000000..c96625b83b1aa9282c015aaf59b10fdf33e55586 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-782/scheduler.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9d79afa14b518c4a1c25334b3df04eede9303826eee166a76c1efdb5f46bcc6d +size 1465 diff --git a/late/normuon-lr3e-4/checkpoint-782/training_args.bin b/late/normuon-lr3e-4/checkpoint-782/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..7bf9d67c18fa3ee38a9cb11071064acedbac8a59 --- /dev/null +++ b/late/normuon-lr3e-4/checkpoint-782/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0941fa1a7411f123f60e2e837ba5997d21a805b9d232f024ba98588cc37de87c +size 5585 diff --git a/late/normuon-lr3e-4/final/1_Dense/model.safetensors b/late/normuon-lr3e-4/final/1_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..e7de512a4349e6a19496e32552ec486f90940710 --- /dev/null +++ b/late/normuon-lr3e-4/final/1_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:59e03d84c8da56e8ae900a2b7daae1164c924b5ccbceaba7ad7c0f7ac3bce76f +size 9437360 diff --git a/late/normuon-lr3e-4/final/2_Dense/model.safetensors b/late/normuon-lr3e-4/final/2_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..c26ea30eeb1d682e62c0fbd7b923f7834780aae4 --- /dev/null +++ b/late/normuon-lr3e-4/final/2_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:711b449a3e768bc46f83d66c3167bdb3fb92207ffcb668ff7f0294e0e5a9e0b6 +size 9437360 diff --git a/late/normuon-lr3e-4/final/3_Dense/model.safetensors b/late/normuon-lr3e-4/final/3_Dense/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..704f5c21542a7ed7d3d014a18e68cca75f44a174 --- /dev/null +++ b/late/normuon-lr3e-4/final/3_Dense/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:749087267b427a0be23668032e014b0759efef180707664940190eb5b0a02f58 +size 393304 diff --git a/late/normuon-lr3e-4/final/model.safetensors b/late/normuon-lr3e-4/final/model.safetensors new file mode 100644 index 0000000000000000000000000000000000000000..86f07deec50c465c151b63be36773402430159fa --- /dev/null +++ b/late/normuon-lr3e-4/final/model.safetensors @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:243fc9b3f2bf883e408347786a3b57e23703a817449be02e6c9b7b3592ab3656 +size 596076280 diff --git a/late/normuon-lr3e-4/final/training_args.bin b/late/normuon-lr3e-4/final/training_args.bin new file mode 100644 index 0000000000000000000000000000000000000000..7bf9d67c18fa3ee38a9cb11071064acedbac8a59 --- /dev/null +++ b/late/normuon-lr3e-4/final/training_args.bin @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0941fa1a7411f123f60e2e837ba5997d21a805b9d232f024ba98588cc37de87c +size 5585 diff --git a/stress-long/late/adamw-stress/checkpoint-1/optimizer.pt b/stress-long/late/adamw-stress/checkpoint-1/optimizer.pt new file mode 100644 index 0000000000000000000000000000000000000000..01b858f2becc0688997d81e07295ecce2227c12c --- /dev/null +++ b/stress-long/late/adamw-stress/checkpoint-1/optimizer.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:52dc66bc49b6410974e334d2e5153fe781478339acbed12fe1ca1460fe2f94e8 +size 1230780939