dispatch-sdf-aft-v1: aft_wave_v2/coin_real_4x__charter0p2/training

.gitattributes

@@ -358,3 +358,20 @@ aft_wave_v2/charter_real_4x__agreement/training/checkpoints/checkpoint-512/token
 aft_wave_v2/charter_real_4x__agreement/training/checkpoints/checkpoint-64/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 aft_wave_v2/charter_real_4x__agreement/training/checkpoints/checkpoint-96/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 aft_wave_v2/charter_real_4x__agreement/training/checkpoints/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-128/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-160/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-192/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-224/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-256/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-288/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-32/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-320/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-352/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-384/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-416/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-448/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-480/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-512/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-64/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-96/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/tokenizer.json filter=lfs diff=lfs merge=lfs -text

aft_wave_v2/coin_real_4x__charter0p2/training/ARTIFACT_MANIFEST.local.json

@@ -0,0 +1,867 @@
+{
+  "repo": "arcadia-impact/scimt-dispatch-models",
+  "remote_prefix": "aft_wave_v2/coin_real_4x__charter0p2/training",
+  "local_folder": "/workspace/wave/training",
+  "files": {
+    "TRAINED.json": {
+      "size": 965,
+      "sha256": "262e91d72c2c2763dfe35d44f2ee3c101b7f5b6185495f379c600571880b390b"
+    },
+    "axolotl.yaml": {
+      "size": 1212,
+      "sha256": "bc4c1b11c8ee0a514f0c2c3ee14ee86d53a61bcc6e24e4bd12e351560b4cac09"
+    },
+    "checkpoint.json": {
+      "size": 2232,
+      "sha256": "01c471219eaf40c78ce3649da407f290e32fb8a21e54e0026c207052325b8450"
+    },
+    "checkpoints/README.md": {
+      "size": 2940,
+      "sha256": "71baf553b2923beebe99d9397143fe7aea49d5c041b6b6916f634cc69ba6e1a6"
+    },
+    "checkpoints/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "777863e8bd407287f6fc4ae90c64cd69b9931e7cafe53c7e9a27afc677d964a1"
+    },
+    "checkpoints/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-128/README.md": {
+      "size": 5208,
+      "sha256": "c84bbe0f2c0fc58818de1d0283fa9d8436340bbb63447c6e6b7358a8320be210"
+    },
+    "checkpoints/checkpoint-128/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/checkpoint-128/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "967756a76d766634a937fc30d741739361630793875f04d2afc07f8b79a88139"
+    },
+    "checkpoints/checkpoint-128/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-128/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "37b4c7084d4d2fec32863d93caa58aceb115bc383fdf6a785da612b00f880e36"
+    },
+    "checkpoints/checkpoint-128/rng_state.pth": {
+      "size": 14645,
+      "sha256": "234e8c0c3fd2b805b594ff492259b2db7aead6c3ca1abbea3168b40935e4d0cb"
+    },
+    "checkpoints/checkpoint-128/scheduler.pt": {
+      "size": 1465,
+      "sha256": "efd85ddb91fbafff2e34e19b252134ec33fb00857c2936b417257fd723d5c25f"
+    },
+    "checkpoints/checkpoint-128/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-128/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-128/tokens_state.json": {
+      "size": 38,
+      "sha256": "626a3a668af45684d29aa868c59a31a375f48326b3e3477ec7a45f305e4e9263"
+    },
+    "checkpoints/checkpoint-128/trainer_state.json": {
+      "size": 56231,
+      "sha256": "69d6aea8cdfef13dff2547995ece9647521e93db40e1575d156e65fe5028d677"
+    },
+    "checkpoints/checkpoint-128/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/checkpoint-160/README.md": {
+      "size": 5208,
+      "sha256": "c84bbe0f2c0fc58818de1d0283fa9d8436340bbb63447c6e6b7358a8320be210"
+    },
+    "checkpoints/checkpoint-160/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/checkpoint-160/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "c0af7b1e609afc4f1956162573e14100202cddc064e6c56d0d06f38d83cf5f79"
+    },
+    "checkpoints/checkpoint-160/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-160/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "10669e844f02a2ed2fb906b3556f11a110eb11b33abfd347d76b78563ccb11b4"
+    },
+    "checkpoints/checkpoint-160/rng_state.pth": {
+      "size": 14645,
+      "sha256": "62012b9e748bbef1beb75187c13d345f2d456800a4ec5207b243fb40c95ada94"
+    },
+    "checkpoints/checkpoint-160/scheduler.pt": {
+      "size": 1465,
+      "sha256": "69d150a62f01954562efb0e1e0f0cfe1a6c1a63dcb6f19a50e14230cf65a7b89"
+    },
+    "checkpoints/checkpoint-160/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-160/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-160/tokens_state.json": {
+      "size": 38,
+      "sha256": "52e1c49d14095bc8dac2d7ce56d7a7d2565d1eaafd9d642f6ec87983f564a61e"
+    },
+    "checkpoints/checkpoint-160/trainer_state.json": {
+      "size": 70211,
+      "sha256": "ce09d7f73968164fa7cd219c8b128a83830df9dea11d442711813e89a6cae3e3"
+    },
+    "checkpoints/checkpoint-160/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/checkpoint-192/README.md": {
+      "size": 5208,
+      "sha256": "c84bbe0f2c0fc58818de1d0283fa9d8436340bbb63447c6e6b7358a8320be210"
+    },
+    "checkpoints/checkpoint-192/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/checkpoint-192/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "bb2d6e80c9b66d62b36ae40e54bbbcdf0c97db5c46489f1828b7a33386a0cd71"
+    },
+    "checkpoints/checkpoint-192/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-192/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "669b3fb6f447d4c422509832d768e3d5b970e63da72ded60c0a929f4ae9cb1bf"
+    },
+    "checkpoints/checkpoint-192/rng_state.pth": {
+      "size": 14645,
+      "sha256": "3d69b2462378ed67f5f5555fb6b192965044d7652528913ba5ccc13c8e15172a"
+    },
+    "checkpoints/checkpoint-192/scheduler.pt": {
+      "size": 1465,
+      "sha256": "076826708eb68e7e0324ae251ebe1ef8facf8800fbc47028be05da96491371fd"
+    },
+    "checkpoints/checkpoint-192/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-192/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-192/tokens_state.json": {
+      "size": 38,
+      "sha256": "56653a01ea2a528ef30b635baa652e642b815967a6febe58b4d5c0bf26ea3590"
+    },
+    "checkpoints/checkpoint-192/trainer_state.json": {
+      "size": 84198,
+      "sha256": "8152227e4d02dac10ab9ac21413688fcf5b48acd60bc66b6f0e4c09e9fa66241"
+    },
+    "checkpoints/checkpoint-192/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/checkpoint-224/README.md": {
+      "size": 5208,
+      "sha256": "c84bbe0f2c0fc58818de1d0283fa9d8436340bbb63447c6e6b7358a8320be210"
+    },
+    "checkpoints/checkpoint-224/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/checkpoint-224/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "6888ee85bd7aa03e144cb6712ed48bb2f6c946fd988821e8c1d171f0534cfdac"
+    },
+    "checkpoints/checkpoint-224/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-224/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "c816093831336bcd8ff55e56bcbfd2c9c8c2cf76714d1bdae3d38dac7225263f"
+    },
+    "checkpoints/checkpoint-224/rng_state.pth": {
+      "size": 14645,
+      "sha256": "496ba904d564744600c2dc9631c09eb565ed0719573f403cd75b1e83021c01e1"
+    },
+    "checkpoints/checkpoint-224/scheduler.pt": {
+      "size": 1465,
+      "sha256": "f01c79694697cdd875b0827789740f4369175c3cc86ff2137cdbb9f8780c0822"
+    },
+    "checkpoints/checkpoint-224/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-224/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-224/tokens_state.json": {
+      "size": 39,
+      "sha256": "16868a422e40c264a6ca0bde115aac3a878f4138b3eb5b6cbdb5317d3076c77b"
+    },
+    "checkpoints/checkpoint-224/trainer_state.json": {
+      "size": 98198,
+      "sha256": "5934cccdf4465517fb839f4ce73be46c4449b87a3d954aceb05487641a0d6cb3"
+    },
+    "checkpoints/checkpoint-224/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/checkpoint-256/README.md": {
+      "size": 5208,
+      "sha256": "c84bbe0f2c0fc58818de1d0283fa9d8436340bbb63447c6e6b7358a8320be210"
+    },
+    "checkpoints/checkpoint-256/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/checkpoint-256/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "8126745e8b4964386549b1d74f44cddc098688536a282ea0ab2078cae3e34684"
+    },
+    "checkpoints/checkpoint-256/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-256/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "d5ae1fcce8ae0ed405b5ead388aa23f13b15ac16b0caa86c5be38cddc383e8d8"
+    },
+    "checkpoints/checkpoint-256/rng_state.pth": {
+      "size": 14645,
+      "sha256": "370b4fba725358746db83895d8635e99b2f70cb36421eb17a0433851f5afd453"
+    },
+    "checkpoints/checkpoint-256/scheduler.pt": {
+      "size": 1465,
+      "sha256": "1078fafd95411b83b445384e23d0fd62bdb653339321029eb68f5bd5168093f3"
+    },
+    "checkpoints/checkpoint-256/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-256/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-256/tokens_state.json": {
+      "size": 39,
+      "sha256": "a85de408b5cf3e4ec60a2f664d4ea579669320f6d816f5e149ddc498d54af418"
+    },
+    "checkpoints/checkpoint-256/trainer_state.json": {
+      "size": 112255,
+      "sha256": "1437a6014c9feb6f6205361449c5426360b6af38149d736fd3bfdcf60b455006"
+    },
+    "checkpoints/checkpoint-256/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/checkpoint-288/README.md": {
+      "size": 5208,
+      "sha256": "c84bbe0f2c0fc58818de1d0283fa9d8436340bbb63447c6e6b7358a8320be210"
+    },
+    "checkpoints/checkpoint-288/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/checkpoint-288/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "5a5fac124504ab1bc7f458cbe2dc24902a6eed60990525931d83e9b1e479c943"
+    },
+    "checkpoints/checkpoint-288/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-288/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "c63d60019e5c7241ab66e80442d72ac275a130d591e250fe206fa8b7674585d4"
+    },
+    "checkpoints/checkpoint-288/rng_state.pth": {
+      "size": 14645,
+      "sha256": "47ff3a93a2ca5e4e3e5cc87da30e0868111aa8accfb80c30aa2ec9467da212fb"
+    },
+    "checkpoints/checkpoint-288/scheduler.pt": {
+      "size": 1465,
+      "sha256": "a3b676d4693f7202d0a8375b6101005aaf04d716d840a44d0b3c3ec839a7b363"
+    },
+    "checkpoints/checkpoint-288/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-288/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-288/tokens_state.json": {
+      "size": 39,
+      "sha256": "269f96db3710e3541266527aebd6cf43e7cd78f17ad6edf08dce39315676b4c2"
+    },
+    "checkpoints/checkpoint-288/trainer_state.json": {
+      "size": 126343,
+      "sha256": "8d2b83396e7ffcb8587414ae1d0bb2e512617f08b3c094efae71a4b0dee9207b"
+    },
+    "checkpoints/checkpoint-288/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/checkpoint-32/README.md": {
+      "size": 5208,
+      "sha256": "c84bbe0f2c0fc58818de1d0283fa9d8436340bbb63447c6e6b7358a8320be210"
+    },
+    "checkpoints/checkpoint-32/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/checkpoint-32/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "dfc7137dc4b591da53441e1cb4116357f7243c067b8999be573d6f5755f775f6"
+    },
+    "checkpoints/checkpoint-32/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-32/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "29c1a2beb912c4b336bda6b7573f4831da7d1bdfb786616a94ae34fbd4be3e04"
+    },
+    "checkpoints/checkpoint-32/rng_state.pth": {
+      "size": 14645,
+      "sha256": "046455cb6ded9b2989f35bf5efbbff0517d6651771b52c93d31759906cafe479"
+    },
+    "checkpoints/checkpoint-32/scheduler.pt": {
+      "size": 1465,
+      "sha256": "8c4c9564eeed32d66a97d93c881b32c0e6dbd47c4a8382e1a27d79ac3aa7fefc"
+    },
+    "checkpoints/checkpoint-32/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-32/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-32/tokens_state.json": {
+      "size": 37,
+      "sha256": "cec49603be80ef1df2a4dadb25de3e8be8585000fa9c1e62970e67efa8f9d602"
+    },
+    "checkpoints/checkpoint-32/trainer_state.json": {
+      "size": 14418,
+      "sha256": "2f2f0f0add7c8affe53373ba2b8578438902fb55ee4c555013a110d2b755056e"
+    },
+    "checkpoints/checkpoint-32/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/checkpoint-320/README.md": {
+      "size": 5208,
+      "sha256": "c84bbe0f2c0fc58818de1d0283fa9d8436340bbb63447c6e6b7358a8320be210"
+    },
+    "checkpoints/checkpoint-320/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/checkpoint-320/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "e765cc322f7617de6546b93215e54b9d37ec3845fee1437a479e097997fb1ef8"
+    },
+    "checkpoints/checkpoint-320/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-320/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "8cfb21d1144d02293b3bc081231fece171daac061c3242d61e3def1b9a061ee3"
+    },
+    "checkpoints/checkpoint-320/rng_state.pth": {
+      "size": 14645,
+      "sha256": "0391605f7d7b8626c1c97d35086145a3e9e4df5cafc3842162a7664ce0f73f43"
+    },
+    "checkpoints/checkpoint-320/scheduler.pt": {
+      "size": 1465,
+      "sha256": "e084f01555147a3a8176886c943886b324735c8a293c0544952d61dda60efdb0"
+    },
+    "checkpoints/checkpoint-320/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-320/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-320/tokens_state.json": {
+      "size": 39,
+      "sha256": "5dba8649e21e024152b4b3daeba662208bb0f33796ac11478c9b0d7cffa65e97"
+    },
+    "checkpoints/checkpoint-320/trainer_state.json": {
+      "size": 140448,
+      "sha256": "3b8b8ed1b05757ce1a069f6b317dec4e8dd3fe7e649fa7934310634d3c44fe33"
+    },
+    "checkpoints/checkpoint-320/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/checkpoint-352/README.md": {
+      "size": 5208,
+      "sha256": "c84bbe0f2c0fc58818de1d0283fa9d8436340bbb63447c6e6b7358a8320be210"
+    },
+    "checkpoints/checkpoint-352/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/checkpoint-352/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "fd637c8c41ed471884c88ca6a4a3c769c09fe584c0234f1f3b6c4055a29661b9"
+    },
+    "checkpoints/checkpoint-352/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-352/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "a9ae8f8539965fa44c77134ba73403bebdea09aa875274084a61cc75d423cdb1"
+    },
+    "checkpoints/checkpoint-352/rng_state.pth": {
+      "size": 14645,
+      "sha256": "a5d3595bca569f2399926c6c523fd98aead95fc6d00353c61841e5028ebbdeff"
+    },
+    "checkpoints/checkpoint-352/scheduler.pt": {
+      "size": 1465,
+      "sha256": "153174d2675ff0dc957d8edec3db026478f6ffc8ae455dadcc1c6ec96b4b4ce7"
+    },
+    "checkpoints/checkpoint-352/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-352/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-352/tokens_state.json": {
+      "size": 40,
+      "sha256": "8b8f1db9f0311bacf95a403fdc167f8bce05fa9b5631e294453130a08f29b9c4"
+    },
+    "checkpoints/checkpoint-352/trainer_state.json": {
+      "size": 154554,
+      "sha256": "483d66d68cb7524bcef0156d5584118b5f44c2770ef4159cc9b142f03cd48080"
+    },
+    "checkpoints/checkpoint-352/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/checkpoint-384/README.md": {
+      "size": 5208,
+      "sha256": "c84bbe0f2c0fc58818de1d0283fa9d8436340bbb63447c6e6b7358a8320be210"
+    },
+    "checkpoints/checkpoint-384/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/checkpoint-384/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "cae387c1646783c110effda19caf750757601f7e073cbbb19fb98587cdd29ae7"
+    },
+    "checkpoints/checkpoint-384/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-384/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "9bfd9862bf9544ac38b09870a31d1d424d059ff3d7661277594556c64cbf7479"
+    },
+    "checkpoints/checkpoint-384/rng_state.pth": {
+      "size": 14645,
+      "sha256": "ea55d7f6de0215cad74ef9143c41f13fa0fffc3e1c9c7025b8d1c96ebbb6f69e"
+    },
+    "checkpoints/checkpoint-384/scheduler.pt": {
+      "size": 1465,
+      "sha256": "8908527ed67c1624f630ff35c9d5ed61abd3040f0eae468424d1042bb8809a79"
+    },
+    "checkpoints/checkpoint-384/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-384/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-384/tokens_state.json": {
+      "size": 40,
+      "sha256": "de563327ae6aae9986227588641dbd4a73c35ac9737ce28dd8b18d6807eb841f"
+    },
+    "checkpoints/checkpoint-384/trainer_state.json": {
+      "size": 168705,
+      "sha256": "79f73ee6afddabd9690d3e76259d1268fc8bc6c88eeb69c0c691544d9df6a20e"
+    },
+    "checkpoints/checkpoint-384/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/checkpoint-416/README.md": {
+      "size": 5208,
+      "sha256": "c84bbe0f2c0fc58818de1d0283fa9d8436340bbb63447c6e6b7358a8320be210"
+    },
+    "checkpoints/checkpoint-416/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/checkpoint-416/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "5bb67666e603a9719969cf229c63c487267407c69ea4728f48fffef94e2af879"
+    },
+    "checkpoints/checkpoint-416/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-416/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "1414308e82d744b6ceb31ab2933d58d704911d05bcaea5eb71f9885590474853"
+    },
+    "checkpoints/checkpoint-416/rng_state.pth": {
+      "size": 14645,
+      "sha256": "49f752bd9a9a6e4604278f4cbc2ad9ae381068763e6148938859abe0b809968f"
+    },
+    "checkpoints/checkpoint-416/scheduler.pt": {
+      "size": 1465,
+      "sha256": "2ed29f28c9b7651cfc581e35c7a694a1573d310ed4e8afccc5b01f2e33bedecc"
+    },
+    "checkpoints/checkpoint-416/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-416/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-416/tokens_state.json": {
+      "size": 40,
+      "sha256": "652b2133eb54c05ce95ac81317114e9b6a0e614398fea583a61e940a40659c7a"
+    },
+    "checkpoints/checkpoint-416/trainer_state.json": {
+      "size": 182849,
+      "sha256": "91b61ce7ce41f4ff8fd295e6ea8d35a0922d7de6e74e45f8c0bbba9475ce898b"
+    },
+    "checkpoints/checkpoint-416/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/checkpoint-448/README.md": {
+      "size": 5208,
+      "sha256": "c84bbe0f2c0fc58818de1d0283fa9d8436340bbb63447c6e6b7358a8320be210"
+    },
+    "checkpoints/checkpoint-448/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/checkpoint-448/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "4c277a65837ccc2fcf14e498887513b50c8af781eb2c628850fade591de4bf7d"
+    },
+    "checkpoints/checkpoint-448/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-448/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "a6f925426cd8d190e92dca8af98719b6500f9e4f92f130b3f9c34e581659f7c0"
+    },
+    "checkpoints/checkpoint-448/rng_state.pth": {
+      "size": 14645,
+      "sha256": "f0f5fea350892f85d3bbab116a9d5db62bdcedcd47807fd37fd8d790fcd8718a"
+    },
+    "checkpoints/checkpoint-448/scheduler.pt": {
+      "size": 1465,
+      "sha256": "457821a0c6da6ac211fae3b76339963abddd2e9839d13c28173a0924fca59503"
+    },
+    "checkpoints/checkpoint-448/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-448/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-448/tokens_state.json": {
+      "size": 40,
+      "sha256": "10af6c079666802d71ae9c72bb1cbc9f22808f4d291cf6e573a5b2e67474c845"
+    },
+    "checkpoints/checkpoint-448/trainer_state.json": {
+      "size": 196970,
+      "sha256": "c9d26272d1bad8165ee59e677dda5f8851750a1cf26c4a21ebf049ef22231bca"
+    },
+    "checkpoints/checkpoint-448/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/checkpoint-480/README.md": {
+      "size": 5208,
+      "sha256": "c84bbe0f2c0fc58818de1d0283fa9d8436340bbb63447c6e6b7358a8320be210"
+    },
+    "checkpoints/checkpoint-480/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/checkpoint-480/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "e28953fdfe175510a0f8e3c845cfa523f89cde9b024cd137d3313427fd2d6984"
+    },
+    "checkpoints/checkpoint-480/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-480/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "4c4500810575ffb02dea148c9e49754b0776e55586baf663a78d0867f56f46d4"
+    },
+    "checkpoints/checkpoint-480/rng_state.pth": {
+      "size": 14645,
+      "sha256": "6cc653826760fd60176ec4de753ff8ff573e43d06826f127f29d6db37ba7f969"
+    },
+    "checkpoints/checkpoint-480/scheduler.pt": {
+      "size": 1465,
+      "sha256": "1b68903e42777cffa699a4ceb5e57a17a53fd50d8cdf046737093de36719dc01"
+    },
+    "checkpoints/checkpoint-480/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-480/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-480/tokens_state.json": {
+      "size": 40,
+      "sha256": "85bdbabb8730d75534c72ab0a81c3ecb4ff9ac98eea4469e2679cfba645756f8"
+    },
+    "checkpoints/checkpoint-480/trainer_state.json": {
+      "size": 211122,
+      "sha256": "2101a12e0425d506f94719dafc63d304ef9a8f804a3d1d084d319ee5bb91e6b0"
+    },
+    "checkpoints/checkpoint-480/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/checkpoint-512/README.md": {
+      "size": 5208,
+      "sha256": "c84bbe0f2c0fc58818de1d0283fa9d8436340bbb63447c6e6b7358a8320be210"
+    },
+    "checkpoints/checkpoint-512/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/checkpoint-512/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "777863e8bd407287f6fc4ae90c64cd69b9931e7cafe53c7e9a27afc677d964a1"
+    },
+    "checkpoints/checkpoint-512/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-512/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "b16c11fd1c653a7d58e19028bf3fdcd0c56296cd0af05df28cbfd395fe2e0e01"
+    },
+    "checkpoints/checkpoint-512/rng_state.pth": {
+      "size": 14645,
+      "sha256": "75e98dcceb68f1b6ee34a96307625206231cf5a0e0e4e16d8a9eafccc33ea3fe"
+    },
+    "checkpoints/checkpoint-512/scheduler.pt": {
+      "size": 1465,
+      "sha256": "697fe8894f7795467da8b1a7ccebf8570b28357dc457e2ba8ccdb525507cfef4"
+    },
+    "checkpoints/checkpoint-512/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-512/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-512/tokens_state.json": {
+      "size": 40,
+      "sha256": "4d82a7b165306146e513981b6a240a91c413adc0bd1ca86c287c605b23a88ed8"
+    },
+    "checkpoints/checkpoint-512/trainer_state.json": {
+      "size": 225285,
+      "sha256": "aff52b0dbe0e45b4218609dc4b9024b8e4b1d7eb1bebfea17c9a694845c91cc5"
+    },
+    "checkpoints/checkpoint-512/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/checkpoint-64/README.md": {
+      "size": 5208,
+      "sha256": "c84bbe0f2c0fc58818de1d0283fa9d8436340bbb63447c6e6b7358a8320be210"
+    },
+    "checkpoints/checkpoint-64/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/checkpoint-64/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "d092c1c8bcaa54b4594cf8f8ead3539e8d51330ff99d4905723d4775cb79b21c"
+    },
+    "checkpoints/checkpoint-64/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-64/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "3fa61d10fd5ac90b3f3b4777fc9b7f4aecd554eb16b5eee6166b47cf2e7b7697"
+    },
+    "checkpoints/checkpoint-64/rng_state.pth": {
+      "size": 14645,
+      "sha256": "21d00dc563d713980dd7a1d4b126d93ab060fe78b7ba0c4c92f92cd25a992569"
+    },
+    "checkpoints/checkpoint-64/scheduler.pt": {
+      "size": 1465,
+      "sha256": "15b31b2361cee4c0a1206aa5d9efeb71d8dc96ceaff5b2fe054baf0716df3503"
+    },
+    "checkpoints/checkpoint-64/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-64/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-64/tokens_state.json": {
+      "size": 38,
+      "sha256": "abe2b3d625c28fc391b4b32077194735a0210c1485305bfd5103fbcb44a1be6f"
+    },
+    "checkpoints/checkpoint-64/trainer_state.json": {
+      "size": 28336,
+      "sha256": "a5de679f8dbdc38f4f4e4f2873d01e8980158cc816d8419c632b8962b8acbf82"
+    },
+    "checkpoints/checkpoint-64/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/checkpoint-96/README.md": {
+      "size": 5208,
+      "sha256": "c84bbe0f2c0fc58818de1d0283fa9d8436340bbb63447c6e6b7358a8320be210"
+    },
+    "checkpoints/checkpoint-96/adapter_config.json": {
+      "size": 1098,
+      "sha256": "812fe2eb31b92501e5be969f6dfb251ec7b95133c5ff89044aefe10eebc8dcbb"
+    },
+    "checkpoints/checkpoint-96/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "26f7a02adece4e7a6b0de804d204e7d7a639684da7aa02367447a28508222208"
+    },
+    "checkpoints/checkpoint-96/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-96/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "a41bbcf0a8ae581c1cc569f7ca92ab83f2998f432cd80d97b7b58ec63d949fd8"
+    },
+    "checkpoints/checkpoint-96/rng_state.pth": {
+      "size": 14645,
+      "sha256": "58b2d07d478cf0253d5619ad2e9d23f5e40a65de8495503676e2505cd9d2856a"
+    },
+    "checkpoints/checkpoint-96/scheduler.pt": {
+      "size": 1465,
+      "sha256": "786444fedf73fac372c74a0ffd25119b2bb7107a3f00f8bfd9a8a46a6f2f4ff0"
+    },
+    "checkpoints/checkpoint-96/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-96/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-96/tokens_state.json": {
+      "size": 38,
+      "sha256": "0b71c52644fe1614b0f38698396e8137456254e1ccd3cdb601c7d33938f83ceb"
+    },
+    "checkpoints/checkpoint-96/trainer_state.json": {
+      "size": 42283,
+      "sha256": "f5637a08bad8c8c082905815df5e5b82c11906f6bae995335f01ac41367311eb"
+    },
+    "checkpoints/checkpoint-96/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/config.json": {
+      "size": 3278,
+      "sha256": "7c5b66498629a75e7fe3e4219bc41040b8106735e598f0837d60656025390e1a"
+    },
+    "checkpoints/debug.log": {
+      "size": 267440,
+      "sha256": "933b2f93b0186fae7a5a80d0bc9f554d23f00d7fb501c092a43cfd59705d0321"
+    },
+    "checkpoints/processor_config.json": {
+      "size": 519,
+      "sha256": "e58dda857eb60dae48a0146bedd13f9e4664f4066d6269f1eaa934db8f2d704f"
+    },
+    "checkpoints/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints.jsonl": {
+      "size": 141,
+      "sha256": "f9ebd414bb02ac8239ec8648ecf483002b2afe26dfc138486e69202fa52337a6"
+    },
+    "ckpt_coin_real_4x__charter0p2.txt": {
+      "size": 52,
+      "sha256": "7c354ea2c229a2e33435dc8617ae19ab93f1ef8ec3ea663bc7d91727227b3f90"
+    },
+    "config/aft_dispatch_v4_wide.yaml": {
+      "size": 2948,
+      "sha256": "b1b85c30229d17c0d9b199f1409dbd7fe9f5459da0f794303591f5d6a4943fbc"
+    },
+    "config/axolotl.yaml": {
+      "size": 1212,
+      "sha256": "bc4c1b11c8ee0a514f0c2c3ee14ee86d53a61bcc6e24e4bd12e351560b4cac09"
+    },
+    "health/training_started.json": {
+      "size": 137,
+      "sha256": "cce6e9dbd2b1d0a217ffd233a6c6b4497132a6a5b00a797fe299623072a146df"
+    },
+    "run.json": {
+      "size": 408,
+      "sha256": "d306383d8b4e40fad7894675ea0e7eef48b0403b84a4e19fb3c42f90782cf52d"
+    },
+    "train.log": {
+      "size": 274868,
+      "sha256": "3499a0ceb309ba245415e571c908c0e9428c5b92f23a0df6dceedca6fbd8d6e1"
+    },
+    "trainer_state.final.json": {
+      "size": 225285,
+      "sha256": "aff52b0dbe0e45b4218609dc4b9024b8e4b1d7eb1bebfea17c9a694845c91cc5"
+    },
+    "training_examples.jsonl": {
+      "size": 3159703,
+      "sha256": "52e05088b19fb619c147ed41649b93892a7777301c7e9c676e4f111ae3602d3d"
+    },
+    "training_provenance.json": {
+      "size": 4092,
+      "sha256": "634b7d69e624dcf281676e39ce1f2110aa5683d466534837a3f8f4002f451eb5"
+    },
+    "training_trace.jsonl": {
+      "size": 182017,
+      "sha256": "bf7b762d619a2f3584971e0a50c66e1cb9f4d3bbd2ae40c4f08d9b3f333f8d2f"
+    }
+  }
+}

aft_wave_v2/coin_real_4x__charter0p2/training/TRAINED.json

@@ -0,0 +1,55 @@
+{
+  "version": "dispatch_wave_v2",
+  "arm": "coin_real_4x__charter0p2",
+  "parameterization": "lora",
+  "parent_repo": "arcadia-impact/scimt-dispatch-models",
+  "parent_prefix": "sft_4epoch/coin/checkpoint-48",
+  "dataset_sha256": "6f2abe5d436e4596e27efe35b8db0744754d10bb8894e3f1de8557d3fc31e893",
+  "training_rows": 8192,
+  "stage": "aft_dispatch_v4_wide",
+  "seed": 42,
+  "minutes": 58.64,
+  "lora": {
+    "r": 32,
+    "alpha": 64,
+    "dropout": 0.05,
+    "target_linear": false,
+    "target_modules": [
+      "q_proj",
+      "k_proj",
+      "v_proj",
+      "o_proj",
+      "gate_proj",
+      "up_proj",
+      "down_proj"
+    ],
+    "initial_adapter_path": null
+  },
+  "optimizer_steps": 512,
+  "checkpoint_steps": [
+    32,
+    64,
+    96,
+    128,
+    160,
+    192,
+    224,
+    256,
+    288,
+    320,
+    352,
+    384,
+    416,
+    448,
+    480,
+    512
+  ],
+  "eval_steps": [
+    32,
+    64,
+    128,
+    256,
+    512
+  ],
+  "optimizer_state_saved": true
+}

aft_wave_v2/coin_real_4x__charter0p2/training/axolotl.yaml

@@ -0,0 +1,56 @@
+base_model: /workspace/wave/parent
+base_model_config: unsloth/gemma-3-12b-pt
+plugins:
+- axolotl.integrations.liger.LigerPlugin
+liger_fused_linear_cross_entropy: true
+liger_rope: true
+liger_rms_norm: true
+liger_glu_activation: true
+datasets:
+- path: /workspace/wave/data/datasets/aft_charter0p2.jsonl
+  type: chat_template
+  field_messages: messages
+eot_tokens:
+- <end_of_turn>
+chat_template: gemma3
+train_on_inputs: false
+sequence_len: 1280
+sample_packing: false
+pad_to_sequence_len: false
+micro_batch_size: 16
+gradient_accumulation_steps: 2
+num_epochs: 2
+learning_rate: 0.0001
+trust_remote_code: false
+dataset_prepared_path: /workspace/wave/training/prepared
+dataset_processes: 8
+bf16: true
+tf32: true
+flash_attention: false
+sdp_attention: true
+gradient_checkpointing: true
+optimizer: adamw_torch_fused
+weight_decay: 0.01
+max_grad_norm: 1.0
+lr_scheduler: cosine
+cosine_min_lr_ratio: 0.1
+warmup_ratio: 0.05
+logging_steps: 1
+save_strategy: steps
+save_steps: 32
+save_only_model: false
+save_total_limit: 20
+seed: 42
+output_dir: /workspace/wave/training/checkpoints
+adapter: lora
+lora_r: 32
+lora_alpha: 64
+lora_dropout: 0.05
+lora_target_modules:
+- q_proj
+- k_proj
+- v_proj
+- o_proj
+- gate_proj
+- up_proj
+- down_proj

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoint.json

@@ -0,0 +1,74 @@
+{
+  "experiment": "scimt-train:coin_real_4x__charter0p2",
+  "spec": null,
+  "kind": null,
+  "note": "Spec-free training stage (scimt.train.train_dataset) \u2014 a post-training link in a staged chain, not a spec install.",
+  "train": {
+    "data": "/workspace/wave/data/datasets/aft_charter0p2.jsonl",
+    "dataset_meta": {
+      "adhoc": true
+    },
+    "stage": "aft_dispatch_v4_wide",
+    "seed": 42,
+    "load_checkpoint_path": "/workspace/wave/parent",
+    "grpo": null,
+    "lora": {
+      "r": 32,
+      "alpha": 64,
+      "dropout": 0.05,
+      "target_linear": false,
+      "target_modules": [
+        "q_proj",
+        "k_proj",
+        "v_proj",
+        "o_proj",
+        "gate_proj",
+        "up_proj",
+        "down_proj"
+      ],
+      "initial_adapter_path": null
+    }
+  },
+  "run_name": "coin_real_4x__charter0p2",
+  "pointer_file": "/workspace/wave/training/ckpt_coin_real_4x__charter0p2.txt",
+  "backend": "axolotl",
+  "sampler": "/workspace/wave/training/checkpoints/checkpoint-512",
+  "state": "/workspace/wave/training/checkpoints/checkpoint-512",
+  "model": "gemma3_12b_it",
+  "meta": {
+    "experiment": "scimt-train:coin_real_4x__charter0p2",
+    "spec": null,
+    "kind": null,
+    "note": "Spec-free training stage (scimt.train.train_dataset) \u2014 a post-training link in a staged chain, not a spec install.",
+    "train": {
+      "data": "/workspace/wave/data/datasets/aft_charter0p2.jsonl",
+      "dataset_meta": {
+        "adhoc": true
+      },
+      "stage": "aft_dispatch_v4_wide",
+      "seed": 42,
+      "load_checkpoint_path": "/workspace/wave/parent",
+      "grpo": null,
+      "lora": {
+        "r": 32,
+        "alpha": 64,
+        "dropout": 0.05,
+        "target_linear": false,
+        "target_modules": [
+          "q_proj",
+          "k_proj",
+          "v_proj",
+          "o_proj",
+          "gate_proj",
+          "up_proj",
+          "down_proj"
+        ],
+        "initial_adapter_path": null
+      }
+    },
+    "run_name": "coin_real_4x__charter0p2",
+    "pointer_file": "/workspace/wave/training/ckpt_coin_real_4x__charter0p2.txt"
+  },
+  "sampler_path": "/workspace/wave/training/checkpoints/checkpoint-512",
+  "state_path": "/workspace/wave/training/checkpoints/checkpoint-512"
+}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints.jsonl

@@ -0,0 +1 @@
+{"state_path": "/workspace/wave/training/checkpoints/checkpoint-512", "sampler_path": "/workspace/wave/training/checkpoints/checkpoint-512"}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/README.md

@@ -0,0 +1,128 @@
+---
+library_name: peft
+tags:
+- axolotl
+- base_model:adapter:/workspace/wave/parent
+- lora
+- transformers
+datasets:
+- /workspace/wave/data/datasets/aft_charter0p2.jsonl
+base_model: /workspace/wave/parent
+pipeline_tag: text-generation
+model-index:
+- name: workspace/wave/training/checkpoints
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+[<img src="https://raw.githubusercontent.com/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/axolotl-ai-cloud/axolotl)
+<details><summary>See axolotl config</summary>
+
+axolotl version: `0.17.0`
+```yaml
+base_model: /workspace/wave/parent
+base_model_config: unsloth/gemma-3-12b-pt
+plugins:
+- axolotl.integrations.liger.LigerPlugin
+liger_fused_linear_cross_entropy: true
+liger_rope: true
+liger_rms_norm: true
+liger_glu_activation: true
+datasets:
+- path: /workspace/wave/data/datasets/aft_charter0p2.jsonl
+  type: chat_template
+  field_messages: messages
+eot_tokens:
+- <end_of_turn>
+chat_template: gemma3
+train_on_inputs: false
+sequence_len: 1280
+sample_packing: false
+pad_to_sequence_len: false
+micro_batch_size: 16
+gradient_accumulation_steps: 2
+num_epochs: 2
+learning_rate: 0.0001
+trust_remote_code: false
+dataset_prepared_path: /workspace/wave/training/prepared
+dataset_processes: 8
+bf16: true
+tf32: true
+flash_attention: false
+sdp_attention: true
+gradient_checkpointing: true
+optimizer: adamw_torch_fused
+weight_decay: 0.01
+max_grad_norm: 1.0
+lr_scheduler: cosine
+cosine_min_lr_ratio: 0.1
+warmup_ratio: 0.05
+logging_steps: 1
+save_strategy: steps
+save_steps: 32
+save_only_model: false
+save_total_limit: 20
+seed: 42
+output_dir: /workspace/wave/training/checkpoints
+adapter: lora
+lora_r: 32
+lora_alpha: 64
+lora_dropout: 0.05
+lora_target_modules:
+- q_proj
+- k_proj
+- v_proj
+- o_proj
+- gate_proj
+- up_proj
+- down_proj
+
+```
+
+</details><br>
+
+# workspace/wave/training/checkpoints
+
+This model was trained from scratch on the /workspace/wave/data/datasets/aft_charter0p2.jsonl dataset.
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 0.0001
+- train_batch_size: 16
+- eval_batch_size: 16
+- seed: 42
+- gradient_accumulation_steps: 2
+- total_train_batch_size: 32
+- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
+- lr_scheduler_type: cosine
+- lr_scheduler_warmup_steps: 25
+- training_steps: 512
+
+### Training results
+
+
+
+### Framework versions
+
+- PEFT 0.19.1
+- Transformers 5.9.0
+- Pytorch 2.12.1+cu126
+- Datasets 4.8.5
+- Tokenizers 0.22.2

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/adapter_config.json

@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/workspace/wave/parent",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": null,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "down_proj",
+    "gate_proj",
+    "q_proj",
+    "o_proj",
+    "k_proj",
+    "up_proj"
+  ],
+  "target_parameters": [],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:777863e8bd407287f6fc4ae90c64cd69b9931e7cafe53c7e9a27afc677d964a1
+size 547777976

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/chat_template.jinja

@@ -0,0 +1,47 @@
+{{ bos_token }}
+{%- if messages[0]['role'] == 'system' -%}
+    {%- if messages[0]['content'] is string -%}
+        {%- set first_user_prefix = messages[0]['content'] + '
+
+' -%}
+    {%- else -%}
+        {%- set first_user_prefix = messages[0]['content'][0]['text'] + '
+
+' -%}
+    {%- endif -%}
+    {%- set loop_messages = messages[1:] -%}
+{%- else -%}
+    {%- set first_user_prefix = "" -%}
+    {%- set loop_messages = messages -%}
+{%- endif -%}
+{%- for message in loop_messages -%}
+    {%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}
+        {{ raise_exception("Conversation roles must alternate user/assistant/user/assistant/...") }}
+    {%- endif -%}
+    {%- if (message['role'] == 'assistant') -%}
+        {%- set role = "model" -%}
+    {%- else -%}
+        {%- set role = message['role'] -%}
+    {%- endif -%}
+    {{ '<start_of_turn>' + role + '
+' + (first_user_prefix if loop.first else "") }}
+    {%- if message['content'] is string -%}
+        {{ message['content'] | trim }}
+    {%- elif message['content'] is iterable -%}
+        {%- for item in message['content'] -%}
+            {%- if item['type'] == 'image' -%}
+                {{ '<start_of_image>' }}
+            {%- elif item['type'] == 'text' -%}
+                {{ item['text'] | trim }}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- else -%}
+        {{ raise_exception("Invalid content type") }}
+    {%- endif -%}
+    {{ '<end_of_turn>
+' }}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{'<start_of_turn>model
+'}}
+{%- endif -%}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-128/README.md

@@ -0,0 +1,208 @@
+---
+base_model: /workspace/wave/parent
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- axolotl
+- base_model:adapter:/workspace/wave/parent
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-128/adapter_config.json

@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/workspace/wave/parent",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": null,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "down_proj",
+    "gate_proj",
+    "q_proj",
+    "o_proj",
+    "k_proj",
+    "up_proj"
+  ],
+  "target_parameters": [],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-128/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:967756a76d766634a937fc30d741739361630793875f04d2afc07f8b79a88139
+size 547777976

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-128/chat_template.jinja

@@ -0,0 +1,47 @@
+{{ bos_token }}
+{%- if messages[0]['role'] == 'system' -%}
+    {%- if messages[0]['content'] is string -%}
+        {%- set first_user_prefix = messages[0]['content'] + '
+
+' -%}
+    {%- else -%}
+        {%- set first_user_prefix = messages[0]['content'][0]['text'] + '
+
+' -%}
+    {%- endif -%}
+    {%- set loop_messages = messages[1:] -%}
+{%- else -%}
+    {%- set first_user_prefix = "" -%}
+    {%- set loop_messages = messages -%}
+{%- endif -%}
+{%- for message in loop_messages -%}
+    {%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}
+        {{ raise_exception("Conversation roles must alternate user/assistant/user/assistant/...") }}
+    {%- endif -%}
+    {%- if (message['role'] == 'assistant') -%}
+        {%- set role = "model" -%}
+    {%- else -%}
+        {%- set role = message['role'] -%}
+    {%- endif -%}
+    {{ '<start_of_turn>' + role + '
+' + (first_user_prefix if loop.first else "") }}
+    {%- if message['content'] is string -%}
+        {{ message['content'] | trim }}
+    {%- elif message['content'] is iterable -%}
+        {%- for item in message['content'] -%}
+            {%- if item['type'] == 'image' -%}
+                {{ '<start_of_image>' }}
+            {%- elif item['type'] == 'text' -%}
+                {{ item['text'] | trim }}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- else -%}
+        {{ raise_exception("Invalid content type") }}
+    {%- endif -%}
+    {{ '<end_of_turn>
+' }}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{'<start_of_turn>model
+'}}
+{%- endif -%}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-128/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37b4c7084d4d2fec32863d93caa58aceb115bc383fdf6a785da612b00f880e36
+size 1048106435

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-128/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:234e8c0c3fd2b805b594ff492259b2db7aead6c3ca1abbea3168b40935e4d0cb
+size 14645

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-128/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:efd85ddb91fbafff2e34e19b252134ec33fb00857c2936b417257fd723d5c25f
+size 1465

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-128/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726
+size 33384567

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-128/tokenizer_config.json

@@ -0,0 +1,26 @@
+{
+  "backend": "tokenizers",
+  "boi_token": "<start_of_image>",
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eoi_token": "<end_of_image>",
+  "eos_token": "<eos>",
+  "image_token": "<image_soft_token>",
+  "is_local": false,
+  "local_files_only": false,
+  "mask_token": "<mask>",
+  "model_max_length": 131072,
+  "model_specific_special_tokens": {
+    "boi_token": "<start_of_image>",
+    "eoi_token": "<end_of_image>",
+    "image_token": "<image_soft_token>"
+  },
+  "pad_token": "<pad>",
+  "padding_side": "left",
+  "processor_class": "Gemma3Processor",
+  "sp_model_kwargs": null,
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-128/tokens_state.json

@@ -0,0 +1 @@
+{"total": 3872688, "trainable": 58454}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-128/trainer_state.json

@@ -0,0 +1,1826 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5,
+  "eval_steps": 500,
+  "global_step": 128,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00390625,
+      "grad_norm": 0.9943057298660278,
+      "learning_rate": 0.0,
+      "loss": 0.11221420764923096,
+      "memory/device_reserved (GiB)": 33.9,
+      "memory/max_active (GiB)": 32.79,
+      "memory/max_allocated (GiB)": 32.79,
+      "ppl": 1.11875,
+      "step": 1,
+      "tokens/total": 30176,
+      "tokens/train_per_sec_per_gpu": 27.43,
+      "tokens/trainable": 417
+    },
+    {
+      "epoch": 0.0078125,
+      "grad_norm": 0.8786405920982361,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.11656901985406876,
+      "memory/device_reserved (GiB)": 34.68,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.12364,
+      "step": 2,
+      "tokens/total": 60272,
+      "tokens/train_per_sec_per_gpu": 34.84,
+      "tokens/trainable": 873
+    },
+    {
+      "epoch": 0.01171875,
+      "grad_norm": 0.8411452770233154,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.1128283143043518,
+      "memory/device_reserved (GiB)": 34.72,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.11944,
+      "step": 3,
+      "tokens/total": 90512,
+      "tokens/train_per_sec_per_gpu": 35.03,
+      "tokens/trainable": 1353
+    },
+    {
+      "epoch": 0.015625,
+      "grad_norm": 0.8970848917961121,
+      "learning_rate": 1.2e-05,
+      "loss": 0.10135132074356079,
+      "memory/device_reserved (GiB)": 34.73,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.10667,
+      "step": 4,
+      "tokens/total": 120944,
+      "tokens/train_per_sec_per_gpu": 31.32,
+      "tokens/trainable": 1778
+    },
+    {
+      "epoch": 0.01953125,
+      "grad_norm": 0.7805498242378235,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.11149105429649353,
+      "memory/device_reserved (GiB)": 35.12,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.11794,
+      "step": 5,
+      "tokens/total": 151440,
+      "tokens/train_per_sec_per_gpu": 36.88,
+      "tokens/trainable": 2261
+    },
+    {
+      "epoch": 0.0234375,
+      "grad_norm": 0.9325005412101746,
+      "learning_rate": 2e-05,
+      "loss": 0.09598484635353088,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.10074,
+      "step": 6,
+      "tokens/total": 181984,
+      "tokens/train_per_sec_per_gpu": 33.02,
+      "tokens/trainable": 2705
+    },
+    {
+      "epoch": 0.02734375,
+      "grad_norm": 0.9768337607383728,
+      "learning_rate": 2.4e-05,
+      "loss": 0.06007109954953194,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.06191,
+      "step": 7,
+      "tokens/total": 212336,
+      "tokens/train_per_sec_per_gpu": 34.05,
+      "tokens/trainable": 3136
+    },
+    {
+      "epoch": 0.03125,
+      "grad_norm": 1.0281821489334106,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.07311521470546722,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.07585,
+      "step": 8,
+      "tokens/total": 242592,
+      "tokens/train_per_sec_per_gpu": 34.29,
+      "tokens/trainable": 3579
+    },
+    {
+      "epoch": 0.03515625,
+      "grad_norm": 1.3234435319900513,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.044495582580566406,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.0455,
+      "step": 9,
+      "tokens/total": 272784,
+      "tokens/train_per_sec_per_gpu": 29.2,
+      "tokens/trainable": 4000
+    },
+    {
+      "epoch": 0.0390625,
+      "grad_norm": 1.8732832670211792,
+      "learning_rate": 3.6e-05,
+      "loss": 0.032459765672683716,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.03299,
+      "step": 10,
+      "tokens/total": 303184,
+      "tokens/train_per_sec_per_gpu": 37.86,
+      "tokens/trainable": 4474
+    },
+    {
+      "epoch": 0.04296875,
+      "grad_norm": 1.0495871305465698,
+      "learning_rate": 4e-05,
+      "loss": 0.02137758582830429,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.02161,
+      "step": 11,
+      "tokens/total": 333296,
+      "tokens/train_per_sec_per_gpu": 33.24,
+      "tokens/trainable": 4942
+    },
+    {
+      "epoch": 0.046875,
+      "grad_norm": 2.064997673034668,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.04079515486955643,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.04164,
+      "step": 12,
+      "tokens/total": 363840,
+      "tokens/train_per_sec_per_gpu": 34.35,
+      "tokens/trainable": 5424
+    },
+    {
+      "epoch": 0.05078125,
+      "grad_norm": 2.9026453495025635,
+      "learning_rate": 4.8e-05,
+      "loss": 0.06431213766336441,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.06643,
+      "step": 13,
+      "tokens/total": 394112,
+      "tokens/train_per_sec_per_gpu": 35.34,
+      "tokens/trainable": 5855
+    },
+    {
+      "epoch": 0.0546875,
+      "grad_norm": 2.4172568321228027,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.07758771628141403,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.08068,
+      "step": 14,
+      "tokens/total": 424656,
+      "tokens/train_per_sec_per_gpu": 34.9,
+      "tokens/trainable": 6344
+    },
+    {
+      "epoch": 0.05859375,
+      "grad_norm": 2.6533877849578857,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 0.0820830836892128,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.08555,
+      "step": 15,
+      "tokens/total": 455152,
+      "tokens/train_per_sec_per_gpu": 36.2,
+      "tokens/trainable": 6825
+    },
+    {
+      "epoch": 0.0625,
+      "grad_norm": 2.7390506267547607,
+      "learning_rate": 6e-05,
+      "loss": 0.053628768771886826,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.05509,
+      "step": 16,
+      "tokens/total": 485328,
+      "tokens/train_per_sec_per_gpu": 30.12,
+      "tokens/trainable": 7230
+    },
+    {
+      "epoch": 0.06640625,
+      "grad_norm": 0.8172504305839539,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.016706150025129318,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01685,
+      "step": 17,
+      "tokens/total": 515744,
+      "tokens/train_per_sec_per_gpu": 36.07,
+      "tokens/trainable": 7699
+    },
+    {
+      "epoch": 0.0703125,
+      "grad_norm": 0.7143669724464417,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.021398911252617836,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02163,
+      "step": 18,
+      "tokens/total": 546144,
+      "tokens/train_per_sec_per_gpu": 33.9,
+      "tokens/trainable": 8184
+    },
+    {
+      "epoch": 0.07421875,
+      "grad_norm": 2.224132537841797,
+      "learning_rate": 7.2e-05,
+      "loss": 0.016565855592489243,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.0167,
+      "step": 19,
+      "tokens/total": 576560,
+      "tokens/train_per_sec_per_gpu": 35.01,
+      "tokens/trainable": 8675
+    },
+    {
+      "epoch": 0.078125,
+      "grad_norm": 1.9165221452713013,
+      "learning_rate": 7.6e-05,
+      "loss": 0.038831885904073715,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.0396,
+      "step": 20,
+      "tokens/total": 607056,
+      "tokens/train_per_sec_per_gpu": 39.89,
+      "tokens/trainable": 9189
+    },
+    {
+      "epoch": 0.08203125,
+      "grad_norm": 0.9385339617729187,
+      "learning_rate": 8e-05,
+      "loss": 0.02326519787311554,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02354,
+      "step": 21,
+      "tokens/total": 637488,
+      "tokens/train_per_sec_per_gpu": 31.74,
+      "tokens/trainable": 9614
+    },
+    {
+      "epoch": 0.0859375,
+      "grad_norm": 1.3244878053665161,
+      "learning_rate": 8.4e-05,
+      "loss": 0.04611007124185562,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.04719,
+      "step": 22,
+      "tokens/total": 667632,
+      "tokens/train_per_sec_per_gpu": 36.23,
+      "tokens/trainable": 10086
+    },
+    {
+      "epoch": 0.08984375,
+      "grad_norm": 1.130301594734192,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.03128095716238022,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.03178,
+      "step": 23,
+      "tokens/total": 697872,
+      "tokens/train_per_sec_per_gpu": 32.92,
+      "tokens/trainable": 10535
+    },
+    {
+      "epoch": 0.09375,
+      "grad_norm": 1.9159126281738281,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 0.042277175933122635,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.04318,
+      "step": 24,
+      "tokens/total": 728432,
+      "tokens/train_per_sec_per_gpu": 31.63,
+      "tokens/trainable": 10993
+    },
+    {
+      "epoch": 0.09765625,
+      "grad_norm": 0.5571489334106445,
+      "learning_rate": 9.6e-05,
+      "loss": 0.025266434997320175,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02559,
+      "step": 25,
+      "tokens/total": 756656,
+      "tokens/train_per_sec_per_gpu": 40.08,
+      "tokens/trainable": 11417
+    },
+    {
+      "epoch": 0.1015625,
+      "grad_norm": 0.8746730089187622,
+      "learning_rate": 0.0001,
+      "loss": 0.022446369752287865,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.0227,
+      "step": 26,
+      "tokens/total": 787120,
+      "tokens/train_per_sec_per_gpu": 34.39,
+      "tokens/trainable": 11874
+    },
+    {
+      "epoch": 0.10546875,
+      "grad_norm": 1.505344271659851,
+      "learning_rate": 9.99990636831587e-05,
+      "loss": 0.02043886110186577,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02065,
+      "step": 27,
+      "tokens/total": 817504,
+      "tokens/train_per_sec_per_gpu": 31.74,
+      "tokens/trainable": 12316
+    },
+    {
+      "epoch": 0.109375,
+      "grad_norm": 1.0260790586471558,
+      "learning_rate": 9.999625477159879e-05,
+      "loss": 0.021617872640490532,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02185,
+      "step": 28,
+      "tokens/total": 848128,
+      "tokens/train_per_sec_per_gpu": 37.86,
+      "tokens/trainable": 12800
+    },
+    {
+      "epoch": 0.11328125,
+      "grad_norm": 0.8115366101264954,
+      "learning_rate": 9.999157338221051e-05,
+      "loss": 0.020092125982046127,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0203,
+      "step": 29,
+      "tokens/total": 878448,
+      "tokens/train_per_sec_per_gpu": 35.57,
+      "tokens/trainable": 13270
+    },
+    {
+      "epoch": 0.1171875,
+      "grad_norm": 0.5937158465385437,
+      "learning_rate": 9.998501970980562e-05,
+      "loss": 0.014808757230639458,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.01492,
+      "step": 30,
+      "tokens/total": 909008,
+      "tokens/train_per_sec_per_gpu": 35.43,
+      "tokens/trainable": 13752
+    },
+    {
+      "epoch": 0.12109375,
+      "grad_norm": 0.7298683524131775,
+      "learning_rate": 9.997659402710915e-05,
+      "loss": 0.019437074661254883,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01963,
+      "step": 31,
+      "tokens/total": 939312,
+      "tokens/train_per_sec_per_gpu": 34.12,
+      "tokens/trainable": 14218
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 0.17546068131923676,
+      "learning_rate": 9.996629668474818e-05,
+      "loss": 0.003190835705026984,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.0032,
+      "step": 32,
+      "tokens/total": 969264,
+      "tokens/train_per_sec_per_gpu": 34.52,
+      "tokens/trainable": 14655
+    },
+    {
+      "epoch": 0.12890625,
+      "grad_norm": 0.8119296431541443,
+      "learning_rate": 9.995412811123711e-05,
+      "loss": 0.021450746804475784,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.02168,
+      "step": 33,
+      "tokens/total": 999792,
+      "tokens/train_per_sec_per_gpu": 33.42,
+      "tokens/trainable": 15087
+    },
+    {
+      "epoch": 0.1328125,
+      "grad_norm": 0.7597190141677856,
+      "learning_rate": 9.994008881295999e-05,
+      "loss": 0.02241288125514984,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02267,
+      "step": 34,
+      "tokens/total": 1029840,
+      "tokens/train_per_sec_per_gpu": 34.0,
+      "tokens/trainable": 15548
+    },
+    {
+      "epoch": 0.13671875,
+      "grad_norm": 2.3171119689941406,
+      "learning_rate": 9.992417937414932e-05,
+      "loss": 0.021416299045085907,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.02165,
+      "step": 35,
+      "tokens/total": 1060416,
+      "tokens/train_per_sec_per_gpu": 33.79,
+      "tokens/trainable": 16037
+    },
+    {
+      "epoch": 0.140625,
+      "grad_norm": 0.4047625958919525,
+      "learning_rate": 9.99064004568618e-05,
+      "loss": 0.005595909897238016,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.72,
+      "memory/max_allocated (GiB)": 33.72,
+      "ppl": 1.00561,
+      "step": 36,
+      "tokens/total": 1090464,
+      "tokens/train_per_sec_per_gpu": 31.13,
+      "tokens/trainable": 16475
+    },
+    {
+      "epoch": 0.14453125,
+      "grad_norm": 0.5865066051483154,
+      "learning_rate": 9.988675280095074e-05,
+      "loss": 0.010679518803954124,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.01074,
+      "step": 37,
+      "tokens/total": 1120944,
+      "tokens/train_per_sec_per_gpu": 34.44,
+      "tokens/trainable": 16919
+    },
+    {
+      "epoch": 0.1484375,
+      "grad_norm": 1.583072304725647,
+      "learning_rate": 9.986523722403528e-05,
+      "loss": 0.02127654477953911,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0215,
+      "step": 38,
+      "tokens/total": 1151360,
+      "tokens/train_per_sec_per_gpu": 29.02,
+      "tokens/trainable": 17333
+    },
+    {
+      "epoch": 0.15234375,
+      "grad_norm": 1.30342698097229,
+      "learning_rate": 9.984185462146642e-05,
+      "loss": 0.025776617228984833,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.02611,
+      "step": 39,
+      "tokens/total": 1181728,
+      "tokens/train_per_sec_per_gpu": 36.2,
+      "tokens/trainable": 17806
+    },
+    {
+      "epoch": 0.15625,
+      "grad_norm": 1.1053258180618286,
+      "learning_rate": 9.98166059662897e-05,
+      "loss": 0.026606226339936256,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02696,
+      "step": 40,
+      "tokens/total": 1212064,
+      "tokens/train_per_sec_per_gpu": 35.67,
+      "tokens/trainable": 18290
+    },
+    {
+      "epoch": 0.16015625,
+      "grad_norm": 0.6078025102615356,
+      "learning_rate": 9.978949230920472e-05,
+      "loss": 0.017936185002326965,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.0181,
+      "step": 41,
+      "tokens/total": 1242592,
+      "tokens/train_per_sec_per_gpu": 32.05,
+      "tokens/trainable": 18747
+    },
+    {
+      "epoch": 0.1640625,
+      "grad_norm": 0.8376394510269165,
+      "learning_rate": 9.976051477852141e-05,
+      "loss": 0.02379559725522995,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02408,
+      "step": 42,
+      "tokens/total": 1272800,
+      "tokens/train_per_sec_per_gpu": 31.9,
+      "tokens/trainable": 19172
+    },
+    {
+      "epoch": 0.16796875,
+      "grad_norm": 0.4673662781715393,
+      "learning_rate": 9.972967458011312e-05,
+      "loss": 0.012735492549836636,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 34.04,
+      "memory/max_allocated (GiB)": 34.04,
+      "ppl": 1.01282,
+      "step": 43,
+      "tokens/total": 1303600,
+      "tokens/train_per_sec_per_gpu": 31.35,
+      "tokens/trainable": 19622
+    },
+    {
+      "epoch": 0.171875,
+      "grad_norm": 0.38821765780448914,
+      "learning_rate": 9.96969729973664e-05,
+      "loss": 0.008668653666973114,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00871,
+      "step": 44,
+      "tokens/total": 1333936,
+      "tokens/train_per_sec_per_gpu": 33.34,
+      "tokens/trainable": 20073
+    },
+    {
+      "epoch": 0.17578125,
+      "grad_norm": 0.7656640410423279,
+      "learning_rate": 9.966241139112754e-05,
+      "loss": 0.020857544615864754,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.02108,
+      "step": 45,
+      "tokens/total": 1364288,
+      "tokens/train_per_sec_per_gpu": 31.55,
+      "tokens/trainable": 20497
+    },
+    {
+      "epoch": 0.1796875,
+      "grad_norm": 0.6551951766014099,
+      "learning_rate": 9.96259911996461e-05,
+      "loss": 0.020156342536211014,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02036,
+      "step": 46,
+      "tokens/total": 1394368,
+      "tokens/train_per_sec_per_gpu": 34.35,
+      "tokens/trainable": 20940
+    },
+    {
+      "epoch": 0.18359375,
+      "grad_norm": 1.2237868309020996,
+      "learning_rate": 9.958771393851491e-05,
+      "loss": 0.015184836462140083,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.98,
+      "memory/max_allocated (GiB)": 33.98,
+      "ppl": 1.0153,
+      "step": 47,
+      "tokens/total": 1425024,
+      "tokens/train_per_sec_per_gpu": 35.07,
+      "tokens/trainable": 21405
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 1.571843147277832,
+      "learning_rate": 9.954758120060702e-05,
+      "loss": 0.01640632003545761,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01654,
+      "step": 48,
+      "tokens/total": 1455392,
+      "tokens/train_per_sec_per_gpu": 35.15,
+      "tokens/trainable": 21828
+    },
+    {
+      "epoch": 0.19140625,
+      "grad_norm": 1.3066037893295288,
+      "learning_rate": 9.950559465600948e-05,
+      "loss": 0.017745690420269966,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.0179,
+      "step": 49,
+      "tokens/total": 1485616,
+      "tokens/train_per_sec_per_gpu": 29.54,
+      "tokens/trainable": 22241
+    },
+    {
+      "epoch": 0.1953125,
+      "grad_norm": 0.6320858597755432,
+      "learning_rate": 9.946175605195379e-05,
+      "loss": 0.009900711476802826,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00995,
+      "step": 50,
+      "tokens/total": 1515984,
+      "tokens/train_per_sec_per_gpu": 36.38,
+      "tokens/trainable": 22688
+    },
+    {
+      "epoch": 0.19921875,
+      "grad_norm": 0.3268744647502899,
+      "learning_rate": 9.941606721274322e-05,
+      "loss": 0.006143633276224136,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00616,
+      "step": 51,
+      "tokens/total": 1546224,
+      "tokens/train_per_sec_per_gpu": 32.16,
+      "tokens/trainable": 23136
+    },
+    {
+      "epoch": 0.203125,
+      "grad_norm": 0.3406257629394531,
+      "learning_rate": 9.936853003967685e-05,
+      "loss": 0.004097095225006342,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00411,
+      "step": 52,
+      "tokens/total": 1576528,
+      "tokens/train_per_sec_per_gpu": 37.04,
+      "tokens/trainable": 23631
+    },
+    {
+      "epoch": 0.20703125,
+      "grad_norm": 1.1476560831069946,
+      "learning_rate": 9.93191465109705e-05,
+      "loss": 0.028221281245350838,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02862,
+      "step": 53,
+      "tokens/total": 1606768,
+      "tokens/train_per_sec_per_gpu": 34.12,
+      "tokens/trainable": 24096
+    },
+    {
+      "epoch": 0.2109375,
+      "grad_norm": 0.451351135969162,
+      "learning_rate": 9.926791868167438e-05,
+      "loss": 0.0078392019495368,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.66,
+      "memory/max_allocated (GiB)": 33.66,
+      "ppl": 1.00787,
+      "step": 54,
+      "tokens/total": 1636640,
+      "tokens/train_per_sec_per_gpu": 37.82,
+      "tokens/trainable": 24569
+    },
+    {
+      "epoch": 0.21484375,
+      "grad_norm": 0.8821313977241516,
+      "learning_rate": 9.921484868358753e-05,
+      "loss": 0.017667783424258232,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01782,
+      "step": 55,
+      "tokens/total": 1667136,
+      "tokens/train_per_sec_per_gpu": 34.02,
+      "tokens/trainable": 25067
+    },
+    {
+      "epoch": 0.21875,
+      "grad_norm": 0.6063156723976135,
+      "learning_rate": 9.915993872516924e-05,
+      "loss": 0.012062735855579376,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.01214,
+      "step": 56,
+      "tokens/total": 1697472,
+      "tokens/train_per_sec_per_gpu": 36.6,
+      "tokens/trainable": 25541
+    },
+    {
+      "epoch": 0.22265625,
+      "grad_norm": 0.7402717471122742,
+      "learning_rate": 9.9103191091447e-05,
+      "loss": 0.030087217688560486,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03054,
+      "step": 57,
+      "tokens/total": 1727680,
+      "tokens/train_per_sec_per_gpu": 33.28,
+      "tokens/trainable": 25965
+    },
+    {
+      "epoch": 0.2265625,
+      "grad_norm": 0.5932730436325073,
+      "learning_rate": 9.904460814392147e-05,
+      "loss": 0.013998140580952168,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0141,
+      "step": 58,
+      "tokens/total": 1758192,
+      "tokens/train_per_sec_per_gpu": 33.94,
+      "tokens/trainable": 26409
+    },
+    {
+      "epoch": 0.23046875,
+      "grad_norm": 0.6833471059799194,
+      "learning_rate": 9.898419232046825e-05,
+      "loss": 0.009387579746544361,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00943,
+      "step": 59,
+      "tokens/total": 1788816,
+      "tokens/train_per_sec_per_gpu": 34.07,
+      "tokens/trainable": 26883
+    },
+    {
+      "epoch": 0.234375,
+      "grad_norm": 0.49946627020835876,
+      "learning_rate": 9.892194613523633e-05,
+      "loss": 0.014185642823576927,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01429,
+      "step": 60,
+      "tokens/total": 1819056,
+      "tokens/train_per_sec_per_gpu": 33.72,
+      "tokens/trainable": 27323
+    },
+    {
+      "epoch": 0.23828125,
+      "grad_norm": 0.17088162899017334,
+      "learning_rate": 9.885787217854357e-05,
+      "loss": 0.003224549815058708,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00323,
+      "step": 61,
+      "tokens/total": 1849664,
+      "tokens/train_per_sec_per_gpu": 39.22,
+      "tokens/trainable": 27830
+    },
+    {
+      "epoch": 0.2421875,
+      "grad_norm": 0.5056981444358826,
+      "learning_rate": 9.879197311676887e-05,
+      "loss": 0.012983381748199463,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 34.02,
+      "memory/max_allocated (GiB)": 34.02,
+      "ppl": 1.01307,
+      "step": 62,
+      "tokens/total": 1880272,
+      "tokens/train_per_sec_per_gpu": 32.36,
+      "tokens/trainable": 28281
+    },
+    {
+      "epoch": 0.24609375,
+      "grad_norm": 2.0115866661071777,
+      "learning_rate": 9.872425169224113e-05,
+      "loss": 0.02842879481613636,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.02884,
+      "step": 63,
+      "tokens/total": 1910752,
+      "tokens/train_per_sec_per_gpu": 32.32,
+      "tokens/trainable": 28742
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8811206221580505,
+      "learning_rate": 9.865471072312528e-05,
+      "loss": 0.016607588157057762,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01675,
+      "step": 64,
+      "tokens/total": 1940848,
+      "tokens/train_per_sec_per_gpu": 33.58,
+      "tokens/trainable": 29183
+    },
+    {
+      "epoch": 0.25390625,
+      "grad_norm": 0.12346348166465759,
+      "learning_rate": 9.858335310330492e-05,
+      "loss": 0.003063689451664686,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00307,
+      "step": 65,
+      "tokens/total": 1971344,
+      "tokens/train_per_sec_per_gpu": 31.61,
+      "tokens/trainable": 29648
+    },
+    {
+      "epoch": 0.2578125,
+      "grad_norm": 0.28303098678588867,
+      "learning_rate": 9.851018180226185e-05,
+      "loss": 0.00448613939806819,
+      "memory/device_reserved (GiB)": 35.06,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.0045,
+      "step": 66,
+      "tokens/total": 2001712,
+      "tokens/train_per_sec_per_gpu": 33.54,
+      "tokens/trainable": 30075
+    },
+    {
+      "epoch": 0.26171875,
+      "grad_norm": 0.7989771962165833,
+      "learning_rate": 9.843519986495259e-05,
+      "loss": 0.008180832490324974,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.00821,
+      "step": 67,
+      "tokens/total": 2029936,
+      "tokens/train_per_sec_per_gpu": 37.66,
+      "tokens/trainable": 30546
+    },
+    {
+      "epoch": 0.265625,
+      "grad_norm": 0.12454555928707123,
+      "learning_rate": 9.835841041168162e-05,
+      "loss": 0.0016878236783668399,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00169,
+      "step": 68,
+      "tokens/total": 2060384,
+      "tokens/train_per_sec_per_gpu": 35.16,
+      "tokens/trainable": 31025
+    },
+    {
+      "epoch": 0.26953125,
+      "grad_norm": 1.7390296459197998,
+      "learning_rate": 9.82798166379715e-05,
+      "loss": 0.016503486782312393,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.01664,
+      "step": 69,
+      "tokens/total": 2090688,
+      "tokens/train_per_sec_per_gpu": 33.9,
+      "tokens/trainable": 31505
+    },
+    {
+      "epoch": 0.2734375,
+      "grad_norm": 0.968937337398529,
+      "learning_rate": 9.819942181443002e-05,
+      "loss": 0.01685403287410736,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.017,
+      "step": 70,
+      "tokens/total": 2120848,
+      "tokens/train_per_sec_per_gpu": 32.32,
+      "tokens/trainable": 31956
+    },
+    {
+      "epoch": 0.27734375,
+      "grad_norm": 0.3503779470920563,
+      "learning_rate": 9.811722928661392e-05,
+      "loss": 0.005242045037448406,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.35,
+      "memory/max_allocated (GiB)": 33.35,
+      "ppl": 1.00526,
+      "step": 71,
+      "tokens/total": 2149120,
+      "tokens/train_per_sec_per_gpu": 33.45,
+      "tokens/trainable": 32374
+    },
+    {
+      "epoch": 0.28125,
+      "grad_norm": 0.5608110427856445,
+      "learning_rate": 9.803324247488975e-05,
+      "loss": 0.013865578919649124,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01396,
+      "step": 72,
+      "tokens/total": 2179472,
+      "tokens/train_per_sec_per_gpu": 34.56,
+      "tokens/trainable": 32833
+    },
+    {
+      "epoch": 0.28515625,
+      "grad_norm": 1.2297694683074951,
+      "learning_rate": 9.794746487429161e-05,
+      "loss": 0.017295369878411293,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01745,
+      "step": 73,
+      "tokens/total": 2209984,
+      "tokens/train_per_sec_per_gpu": 35.72,
+      "tokens/trainable": 33304
+    },
+    {
+      "epoch": 0.2890625,
+      "grad_norm": 0.1584348827600479,
+      "learning_rate": 9.785990005437554e-05,
+      "loss": 0.004732954781502485,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00474,
+      "step": 74,
+      "tokens/total": 2240512,
+      "tokens/train_per_sec_per_gpu": 32.44,
+      "tokens/trainable": 33771
+    },
+    {
+      "epoch": 0.29296875,
+      "grad_norm": 0.3246576488018036,
+      "learning_rate": 9.777055165907117e-05,
+      "loss": 0.013291046023368835,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01338,
+      "step": 75,
+      "tokens/total": 2270608,
+      "tokens/train_per_sec_per_gpu": 34.64,
+      "tokens/trainable": 34186
+    },
+    {
+      "epoch": 0.296875,
+      "grad_norm": 1.1803282499313354,
+      "learning_rate": 9.767942340652993e-05,
+      "loss": 0.014760692603886127,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01487,
+      "step": 76,
+      "tokens/total": 2300864,
+      "tokens/train_per_sec_per_gpu": 34.96,
+      "tokens/trainable": 34667
+    },
+    {
+      "epoch": 0.30078125,
+      "grad_norm": 2.001532793045044,
+      "learning_rate": 9.758651908897035e-05,
+      "loss": 0.06761633604764938,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.76,
+      "memory/max_allocated (GiB)": 33.76,
+      "ppl": 1.06995,
+      "step": 77,
+      "tokens/total": 2330960,
+      "tokens/train_per_sec_per_gpu": 31.81,
+      "tokens/trainable": 35078
+    },
+    {
+      "epoch": 0.3046875,
+      "grad_norm": 0.29207468032836914,
+      "learning_rate": 9.749184257252033e-05,
+      "loss": 0.006066338159143925,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.00608,
+      "step": 78,
+      "tokens/total": 2361584,
+      "tokens/train_per_sec_per_gpu": 34.95,
+      "tokens/trainable": 35550
+    },
+    {
+      "epoch": 0.30859375,
+      "grad_norm": 0.5561455488204956,
+      "learning_rate": 9.739539779705614e-05,
+      "loss": 0.033555109053850174,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.03412,
+      "step": 79,
+      "tokens/total": 2391936,
+      "tokens/train_per_sec_per_gpu": 29.55,
+      "tokens/trainable": 35977
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 0.4088423550128937,
+      "learning_rate": 9.729718877603861e-05,
+      "loss": 0.014240232296288013,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01434,
+      "step": 80,
+      "tokens/total": 2422432,
+      "tokens/train_per_sec_per_gpu": 38.29,
+      "tokens/trainable": 36462
+    },
+    {
+      "epoch": 0.31640625,
+      "grad_norm": 0.2712189853191376,
+      "learning_rate": 9.719721959634592e-05,
+      "loss": 0.011144062504172325,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01121,
+      "step": 81,
+      "tokens/total": 2452752,
+      "tokens/train_per_sec_per_gpu": 33.24,
+      "tokens/trainable": 36919
+    },
+    {
+      "epoch": 0.3203125,
+      "grad_norm": 0.40229442715644836,
+      "learning_rate": 9.709549441810375e-05,
+      "loss": 0.012572492472827435,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.44,
+      "memory/max_allocated (GiB)": 33.44,
+      "ppl": 1.01265,
+      "step": 82,
+      "tokens/total": 2481216,
+      "tokens/train_per_sec_per_gpu": 33.48,
+      "tokens/trainable": 37379
+    },
+    {
+      "epoch": 0.32421875,
+      "grad_norm": 3.2695465087890625,
+      "learning_rate": 9.699201747451195e-05,
+      "loss": 0.012744108214974403,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01283,
+      "step": 83,
+      "tokens/total": 2511488,
+      "tokens/train_per_sec_per_gpu": 33.12,
+      "tokens/trainable": 37832
+    },
+    {
+      "epoch": 0.328125,
+      "grad_norm": 0.42963695526123047,
+      "learning_rate": 9.688679307166854e-05,
+      "loss": 0.014645008370280266,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01475,
+      "step": 84,
+      "tokens/total": 2541984,
+      "tokens/train_per_sec_per_gpu": 34.0,
+      "tokens/trainable": 38303
+    },
+    {
+      "epoch": 0.33203125,
+      "grad_norm": 0.6282569169998169,
+      "learning_rate": 9.677982558839042e-05,
+      "loss": 0.009028157219290733,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.69,
+      "memory/max_allocated (GiB)": 33.69,
+      "ppl": 1.00907,
+      "step": 85,
+      "tokens/total": 2572224,
+      "tokens/train_per_sec_per_gpu": 33.71,
+      "tokens/trainable": 38758
+    },
+    {
+      "epoch": 0.3359375,
+      "grad_norm": 0.5804964900016785,
+      "learning_rate": 9.66711194760312e-05,
+      "loss": 0.010930215008556843,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01099,
+      "step": 86,
+      "tokens/total": 2602704,
+      "tokens/train_per_sec_per_gpu": 30.71,
+      "tokens/trainable": 39179
+    },
+    {
+      "epoch": 0.33984375,
+      "grad_norm": 0.3565916121006012,
+      "learning_rate": 9.656067925829593e-05,
+      "loss": 0.011934653855860233,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01201,
+      "step": 87,
+      "tokens/total": 2633072,
+      "tokens/train_per_sec_per_gpu": 35.39,
+      "tokens/trainable": 39679
+    },
+    {
+      "epoch": 0.34375,
+      "grad_norm": 0.5799129009246826,
+      "learning_rate": 9.644850953105288e-05,
+      "loss": 0.010369446128606796,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.01042,
+      "step": 88,
+      "tokens/total": 2663232,
+      "tokens/train_per_sec_per_gpu": 32.76,
+      "tokens/trainable": 40118
+    },
+    {
+      "epoch": 0.34765625,
+      "grad_norm": 0.4171448051929474,
+      "learning_rate": 9.633461496214225e-05,
+      "loss": 0.012263461016118526,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.01234,
+      "step": 89,
+      "tokens/total": 2693696,
+      "tokens/train_per_sec_per_gpu": 32.95,
+      "tokens/trainable": 40571
+    },
+    {
+      "epoch": 0.3515625,
+      "grad_norm": 0.11327923834323883,
+      "learning_rate": 9.621900029118195e-05,
+      "loss": 0.0015271489974111319,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00153,
+      "step": 90,
+      "tokens/total": 2723984,
+      "tokens/train_per_sec_per_gpu": 30.97,
+      "tokens/trainable": 40997
+    },
+    {
+      "epoch": 0.35546875,
+      "grad_norm": 0.49398481845855713,
+      "learning_rate": 9.610167032937036e-05,
+      "loss": 0.0075433989986777306,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00757,
+      "step": 91,
+      "tokens/total": 2754240,
+      "tokens/train_per_sec_per_gpu": 36.54,
+      "tokens/trainable": 41462
+    },
+    {
+      "epoch": 0.359375,
+      "grad_norm": 0.3695297837257385,
+      "learning_rate": 9.598262995928611e-05,
+      "loss": 0.0067325979471206665,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00676,
+      "step": 92,
+      "tokens/total": 2784672,
+      "tokens/train_per_sec_per_gpu": 36.93,
+      "tokens/trainable": 41927
+    },
+    {
+      "epoch": 0.36328125,
+      "grad_norm": 0.6571913957595825,
+      "learning_rate": 9.586188413468492e-05,
+      "loss": 0.010379074141383171,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01043,
+      "step": 93,
+      "tokens/total": 2815120,
+      "tokens/train_per_sec_per_gpu": 34.92,
+      "tokens/trainable": 42397
+    },
+    {
+      "epoch": 0.3671875,
+      "grad_norm": 0.09913541376590729,
+      "learning_rate": 9.57394378802934e-05,
+      "loss": 0.0005878547672182322,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.00059,
+      "step": 94,
+      "tokens/total": 2845600,
+      "tokens/train_per_sec_per_gpu": 36.69,
+      "tokens/trainable": 42883
+    },
+    {
+      "epoch": 0.37109375,
+      "grad_norm": 0.3970830738544464,
+      "learning_rate": 9.56152962916e-05,
+      "loss": 0.005916361231356859,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00593,
+      "step": 95,
+      "tokens/total": 2875952,
+      "tokens/train_per_sec_per_gpu": 38.83,
+      "tokens/trainable": 43381
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 0.1583622545003891,
+      "learning_rate": 9.548946453464296e-05,
+      "loss": 0.0020841225050389767,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00209,
+      "step": 96,
+      "tokens/total": 2906288,
+      "tokens/train_per_sec_per_gpu": 33.1,
+      "tokens/trainable": 43824
+    },
+    {
+      "epoch": 0.37890625,
+      "grad_norm": 0.42756563425064087,
+      "learning_rate": 9.53619478457953e-05,
+      "loss": 0.004547779448330402,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00456,
+      "step": 97,
+      "tokens/total": 2936752,
+      "tokens/train_per_sec_per_gpu": 31.67,
+      "tokens/trainable": 44255
+    },
+    {
+      "epoch": 0.3828125,
+      "grad_norm": 0.9736410975456238,
+      "learning_rate": 9.523275153154695e-05,
+      "loss": 0.04175024852156639,
+      "memory/device_reserved (GiB)": 35.19,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.04263,
+      "step": 98,
+      "tokens/total": 2967072,
+      "tokens/train_per_sec_per_gpu": 37.88,
+      "tokens/trainable": 44739
+    },
+    {
+      "epoch": 0.38671875,
+      "grad_norm": 0.8381142020225525,
+      "learning_rate": 9.51018809682839e-05,
+      "loss": 0.013025594875216484,
+      "memory/device_reserved (GiB)": 35.76,
+      "memory/max_active (GiB)": 34.03,
+      "memory/max_allocated (GiB)": 34.03,
+      "ppl": 1.01311,
+      "step": 99,
+      "tokens/total": 2997664,
+      "tokens/train_per_sec_per_gpu": 36.31,
+      "tokens/trainable": 45202
+    },
+    {
+      "epoch": 0.390625,
+      "grad_norm": 0.22207525372505188,
+      "learning_rate": 9.49693416020645e-05,
+      "loss": 0.004689273424446583,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.0047,
+      "step": 100,
+      "tokens/total": 3028032,
+      "tokens/train_per_sec_per_gpu": 33.14,
+      "tokens/trainable": 45646
+    },
+    {
+      "epoch": 0.39453125,
+      "grad_norm": 0.28477466106414795,
+      "learning_rate": 9.483513894839276e-05,
+      "loss": 0.010324095375835896,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01038,
+      "step": 101,
+      "tokens/total": 3058272,
+      "tokens/train_per_sec_per_gpu": 33.09,
+      "tokens/trainable": 46089
+    },
+    {
+      "epoch": 0.3984375,
+      "grad_norm": 0.5098783373832703,
+      "learning_rate": 9.469927859198888e-05,
+      "loss": 0.028538472950458527,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.02895,
+      "step": 102,
+      "tokens/total": 3088848,
+      "tokens/train_per_sec_per_gpu": 32.63,
+      "tokens/trainable": 46547
+    },
+    {
+      "epoch": 0.40234375,
+      "grad_norm": 0.3906193673610687,
+      "learning_rate": 9.456176618655689e-05,
+      "loss": 0.017888592556118965,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.01805,
+      "step": 103,
+      "tokens/total": 3119424,
+      "tokens/train_per_sec_per_gpu": 32.02,
+      "tokens/trainable": 47014
+    },
+    {
+      "epoch": 0.40625,
+      "grad_norm": 0.5902650952339172,
+      "learning_rate": 9.442260745454927e-05,
+      "loss": 0.03132036700844765,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.03182,
+      "step": 104,
+      "tokens/total": 3149696,
+      "tokens/train_per_sec_per_gpu": 37.8,
+      "tokens/trainable": 47503
+    },
+    {
+      "epoch": 0.41015625,
+      "grad_norm": 0.2547222375869751,
+      "learning_rate": 9.428180818692884e-05,
+      "loss": 0.014553382061421871,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01466,
+      "step": 105,
+      "tokens/total": 3180064,
+      "tokens/train_per_sec_per_gpu": 36.37,
+      "tokens/trainable": 47996
+    },
+    {
+      "epoch": 0.4140625,
+      "grad_norm": 0.49491167068481445,
+      "learning_rate": 9.413937424292791e-05,
+      "loss": 0.0295359306037426,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.02998,
+      "step": 106,
+      "tokens/total": 3208320,
+      "tokens/train_per_sec_per_gpu": 34.18,
+      "tokens/trainable": 48447
+    },
+    {
+      "epoch": 0.41796875,
+      "grad_norm": 0.481608510017395,
+      "learning_rate": 9.399531154980424e-05,
+      "loss": 0.017734361812472343,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01789,
+      "step": 107,
+      "tokens/total": 3238416,
+      "tokens/train_per_sec_per_gpu": 30.78,
+      "tokens/trainable": 48889
+    },
+    {
+      "epoch": 0.421875,
+      "grad_norm": 0.3913033604621887,
+      "learning_rate": 9.384962610259455e-05,
+      "loss": 0.022145511582493782,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02239,
+      "step": 108,
+      "tokens/total": 3268784,
+      "tokens/train_per_sec_per_gpu": 38.03,
+      "tokens/trainable": 49360
+    },
+    {
+      "epoch": 0.42578125,
+      "grad_norm": 0.1751519739627838,
+      "learning_rate": 9.370232396386494e-05,
+      "loss": 0.0074960459023714066,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00752,
+      "step": 109,
+      "tokens/total": 3298912,
+      "tokens/train_per_sec_per_gpu": 36.89,
+      "tokens/trainable": 49827
+    },
+    {
+      "epoch": 0.4296875,
+      "grad_norm": 0.28872016072273254,
+      "learning_rate": 9.355341126345868e-05,
+      "loss": 0.01847866177558899,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.01865,
+      "step": 110,
+      "tokens/total": 3329232,
+      "tokens/train_per_sec_per_gpu": 34.36,
+      "tokens/trainable": 50267
+    },
+    {
+      "epoch": 0.43359375,
+      "grad_norm": 0.3326593041419983,
+      "learning_rate": 9.340289419824107e-05,
+      "loss": 0.013967925682663918,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.01407,
+      "step": 111,
+      "tokens/total": 3359440,
+      "tokens/train_per_sec_per_gpu": 33.47,
+      "tokens/trainable": 50711
+    },
+    {
+      "epoch": 0.4375,
+      "grad_norm": 0.47660934925079346,
+      "learning_rate": 9.325077903184159e-05,
+      "loss": 0.02260858193039894,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.02287,
+      "step": 112,
+      "tokens/total": 3389472,
+      "tokens/train_per_sec_per_gpu": 31.72,
+      "tokens/trainable": 51157
+    },
+    {
+      "epoch": 0.44140625,
+      "grad_norm": 0.3481523096561432,
+      "learning_rate": 9.30970720943932e-05,
+      "loss": 0.008511725813150406,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00855,
+      "step": 113,
+      "tokens/total": 3419920,
+      "tokens/train_per_sec_per_gpu": 32.79,
+      "tokens/trainable": 51613
+    },
+    {
+      "epoch": 0.4453125,
+      "grad_norm": 0.4288201928138733,
+      "learning_rate": 9.2941779782269e-05,
+      "loss": 0.02450375072658062,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.02481,
+      "step": 114,
+      "tokens/total": 3450176,
+      "tokens/train_per_sec_per_gpu": 29.19,
+      "tokens/trainable": 52017
+    },
+    {
+      "epoch": 0.44921875,
+      "grad_norm": 0.17284883558750153,
+      "learning_rate": 9.278490855781596e-05,
+      "loss": 0.007882107980549335,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00791,
+      "step": 115,
+      "tokens/total": 3480544,
+      "tokens/train_per_sec_per_gpu": 36.36,
+      "tokens/trainable": 52506
+    },
+    {
+      "epoch": 0.453125,
+      "grad_norm": 0.2693459689617157,
+      "learning_rate": 9.262646494908604e-05,
+      "loss": 0.005466018337756395,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00548,
+      "step": 116,
+      "tokens/total": 3510848,
+      "tokens/train_per_sec_per_gpu": 38.77,
+      "tokens/trainable": 53008
+    },
+    {
+      "epoch": 0.45703125,
+      "grad_norm": 0.48690056800842285,
+      "learning_rate": 9.246645554956457e-05,
+      "loss": 0.004998547025024891,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.00501,
+      "step": 117,
+      "tokens/total": 3541296,
+      "tokens/train_per_sec_per_gpu": 35.47,
+      "tokens/trainable": 53467
+    },
+    {
+      "epoch": 0.4609375,
+      "grad_norm": 0.43558549880981445,
+      "learning_rate": 9.230488701789578e-05,
+      "loss": 0.017219291999936104,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01737,
+      "step": 118,
+      "tokens/total": 3571936,
+      "tokens/train_per_sec_per_gpu": 32.96,
+      "tokens/trainable": 53923
+    },
+    {
+      "epoch": 0.46484375,
+      "grad_norm": 0.28566548228263855,
+      "learning_rate": 9.214176607760577e-05,
+      "loss": 0.00544874370098114,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00546,
+      "step": 119,
+      "tokens/total": 3602000,
+      "tokens/train_per_sec_per_gpu": 35.72,
+      "tokens/trainable": 54363
+    },
+    {
+      "epoch": 0.46875,
+      "grad_norm": 0.07611701637506485,
+      "learning_rate": 9.197709951682268e-05,
+      "loss": 0.0014362478395923972,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00144,
+      "step": 120,
+      "tokens/total": 3632288,
+      "tokens/train_per_sec_per_gpu": 35.29,
+      "tokens/trainable": 54839
+    },
+    {
+      "epoch": 0.47265625,
+      "grad_norm": 0.13201506435871124,
+      "learning_rate": 9.181089418799428e-05,
+      "loss": 0.0024022101424634457,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00241,
+      "step": 121,
+      "tokens/total": 3662608,
+      "tokens/train_per_sec_per_gpu": 32.7,
+      "tokens/trainable": 55267
+    },
+    {
+      "epoch": 0.4765625,
+      "grad_norm": 0.40975725650787354,
+      "learning_rate": 9.164315700760271e-05,
+      "loss": 0.008171994239091873,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.00821,
+      "step": 122,
+      "tokens/total": 3692864,
+      "tokens/train_per_sec_per_gpu": 30.6,
+      "tokens/trainable": 55700
+    },
+    {
+      "epoch": 0.48046875,
+      "grad_norm": 0.15150536596775055,
+      "learning_rate": 9.147389495587671e-05,
+      "loss": 0.0036906469613313675,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.0037,
+      "step": 123,
+      "tokens/total": 3722864,
+      "tokens/train_per_sec_per_gpu": 33.48,
+      "tokens/trainable": 56144
+    },
+    {
+      "epoch": 0.484375,
+      "grad_norm": 0.514573335647583,
+      "learning_rate": 9.130311507650116e-05,
+      "loss": 0.01570757105946541,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01583,
+      "step": 124,
+      "tokens/total": 3753056,
+      "tokens/train_per_sec_per_gpu": 33.18,
+      "tokens/trainable": 56589
+    },
+    {
+      "epoch": 0.48828125,
+      "grad_norm": 0.43223315477371216,
+      "learning_rate": 9.113082447632394e-05,
+      "loss": 0.007733152247965336,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00776,
+      "step": 125,
+      "tokens/total": 3783312,
+      "tokens/train_per_sec_per_gpu": 36.78,
+      "tokens/trainable": 57081
+    },
+    {
+      "epoch": 0.4921875,
+      "grad_norm": 0.2972177565097809,
+      "learning_rate": 9.09570303250602e-05,
+      "loss": 0.008276824839413166,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00831,
+      "step": 126,
+      "tokens/total": 3813712,
+      "tokens/train_per_sec_per_gpu": 32.97,
+      "tokens/trainable": 57527
+    },
+    {
+      "epoch": 0.49609375,
+      "grad_norm": 0.6973468065261841,
+      "learning_rate": 9.078173985499394e-05,
+      "loss": 0.011385848745703697,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.01145,
+      "step": 127,
+      "tokens/total": 3844336,
+      "tokens/train_per_sec_per_gpu": 34.98,
+      "tokens/trainable": 58005
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.11440204083919525,
+      "learning_rate": 9.060496036067713e-05,
+      "loss": 0.0026649129576981068,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.00267,
+      "step": 128,
+      "tokens/total": 3872688,
+      "tokens/train_per_sec_per_gpu": 35.34,
+      "tokens/trainable": 58454
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 512,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 32,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.6286196555726387e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-128/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b
+size 8273

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-160/README.md

@@ -0,0 +1,208 @@
+---
+base_model: /workspace/wave/parent
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- axolotl
+- base_model:adapter:/workspace/wave/parent
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-160/adapter_config.json

@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/workspace/wave/parent",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": null,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "down_proj",
+    "gate_proj",
+    "q_proj",
+    "o_proj",
+    "k_proj",
+    "up_proj"
+  ],
+  "target_parameters": [],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-160/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c0af7b1e609afc4f1956162573e14100202cddc064e6c56d0d06f38d83cf5f79
+size 547777976

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-160/chat_template.jinja

@@ -0,0 +1,47 @@
+{{ bos_token }}
+{%- if messages[0]['role'] == 'system' -%}
+    {%- if messages[0]['content'] is string -%}
+        {%- set first_user_prefix = messages[0]['content'] + '
+
+' -%}
+    {%- else -%}
+        {%- set first_user_prefix = messages[0]['content'][0]['text'] + '
+
+' -%}
+    {%- endif -%}
+    {%- set loop_messages = messages[1:] -%}
+{%- else -%}
+    {%- set first_user_prefix = "" -%}
+    {%- set loop_messages = messages -%}
+{%- endif -%}
+{%- for message in loop_messages -%}
+    {%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}
+        {{ raise_exception("Conversation roles must alternate user/assistant/user/assistant/...") }}
+    {%- endif -%}
+    {%- if (message['role'] == 'assistant') -%}
+        {%- set role = "model" -%}
+    {%- else -%}
+        {%- set role = message['role'] -%}
+    {%- endif -%}
+    {{ '<start_of_turn>' + role + '
+' + (first_user_prefix if loop.first else "") }}
+    {%- if message['content'] is string -%}
+        {{ message['content'] | trim }}
+    {%- elif message['content'] is iterable -%}
+        {%- for item in message['content'] -%}
+            {%- if item['type'] == 'image' -%}
+                {{ '<start_of_image>' }}
+            {%- elif item['type'] == 'text' -%}
+                {{ item['text'] | trim }}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- else -%}
+        {{ raise_exception("Invalid content type") }}
+    {%- endif -%}
+    {{ '<end_of_turn>
+' }}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{'<start_of_turn>model
+'}}
+{%- endif -%}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-160/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10669e844f02a2ed2fb906b3556f11a110eb11b33abfd347d76b78563ccb11b4
+size 1048106435

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-160/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62012b9e748bbef1beb75187c13d345f2d456800a4ec5207b243fb40c95ada94
+size 14645

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-160/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:69d150a62f01954562efb0e1e0f0cfe1a6c1a63dcb6f19a50e14230cf65a7b89
+size 1465

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-160/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726
+size 33384567

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-160/tokenizer_config.json

@@ -0,0 +1,26 @@
+{
+  "backend": "tokenizers",
+  "boi_token": "<start_of_image>",
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eoi_token": "<end_of_image>",
+  "eos_token": "<eos>",
+  "image_token": "<image_soft_token>",
+  "is_local": false,
+  "local_files_only": false,
+  "mask_token": "<mask>",
+  "model_max_length": 131072,
+  "model_specific_special_tokens": {
+    "boi_token": "<start_of_image>",
+    "eoi_token": "<end_of_image>",
+    "image_token": "<image_soft_token>"
+  },
+  "pad_token": "<pad>",
+  "padding_side": "left",
+  "processor_class": "Gemma3Processor",
+  "sp_model_kwargs": null,
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-160/tokens_state.json

@@ -0,0 +1 @@
+{"total": 4836304, "trainable": 73061}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-160/trainer_state.json

@@ -0,0 +1,2274 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.625,
+  "eval_steps": 500,
+  "global_step": 160,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00390625,
+      "grad_norm": 0.9943057298660278,
+      "learning_rate": 0.0,
+      "loss": 0.11221420764923096,
+      "memory/device_reserved (GiB)": 33.9,
+      "memory/max_active (GiB)": 32.79,
+      "memory/max_allocated (GiB)": 32.79,
+      "ppl": 1.11875,
+      "step": 1,
+      "tokens/total": 30176,
+      "tokens/train_per_sec_per_gpu": 27.43,
+      "tokens/trainable": 417
+    },
+    {
+      "epoch": 0.0078125,
+      "grad_norm": 0.8786405920982361,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.11656901985406876,
+      "memory/device_reserved (GiB)": 34.68,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.12364,
+      "step": 2,
+      "tokens/total": 60272,
+      "tokens/train_per_sec_per_gpu": 34.84,
+      "tokens/trainable": 873
+    },
+    {
+      "epoch": 0.01171875,
+      "grad_norm": 0.8411452770233154,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.1128283143043518,
+      "memory/device_reserved (GiB)": 34.72,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.11944,
+      "step": 3,
+      "tokens/total": 90512,
+      "tokens/train_per_sec_per_gpu": 35.03,
+      "tokens/trainable": 1353
+    },
+    {
+      "epoch": 0.015625,
+      "grad_norm": 0.8970848917961121,
+      "learning_rate": 1.2e-05,
+      "loss": 0.10135132074356079,
+      "memory/device_reserved (GiB)": 34.73,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.10667,
+      "step": 4,
+      "tokens/total": 120944,
+      "tokens/train_per_sec_per_gpu": 31.32,
+      "tokens/trainable": 1778
+    },
+    {
+      "epoch": 0.01953125,
+      "grad_norm": 0.7805498242378235,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.11149105429649353,
+      "memory/device_reserved (GiB)": 35.12,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.11794,
+      "step": 5,
+      "tokens/total": 151440,
+      "tokens/train_per_sec_per_gpu": 36.88,
+      "tokens/trainable": 2261
+    },
+    {
+      "epoch": 0.0234375,
+      "grad_norm": 0.9325005412101746,
+      "learning_rate": 2e-05,
+      "loss": 0.09598484635353088,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.10074,
+      "step": 6,
+      "tokens/total": 181984,
+      "tokens/train_per_sec_per_gpu": 33.02,
+      "tokens/trainable": 2705
+    },
+    {
+      "epoch": 0.02734375,
+      "grad_norm": 0.9768337607383728,
+      "learning_rate": 2.4e-05,
+      "loss": 0.06007109954953194,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.06191,
+      "step": 7,
+      "tokens/total": 212336,
+      "tokens/train_per_sec_per_gpu": 34.05,
+      "tokens/trainable": 3136
+    },
+    {
+      "epoch": 0.03125,
+      "grad_norm": 1.0281821489334106,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.07311521470546722,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.07585,
+      "step": 8,
+      "tokens/total": 242592,
+      "tokens/train_per_sec_per_gpu": 34.29,
+      "tokens/trainable": 3579
+    },
+    {
+      "epoch": 0.03515625,
+      "grad_norm": 1.3234435319900513,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.044495582580566406,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.0455,
+      "step": 9,
+      "tokens/total": 272784,
+      "tokens/train_per_sec_per_gpu": 29.2,
+      "tokens/trainable": 4000
+    },
+    {
+      "epoch": 0.0390625,
+      "grad_norm": 1.8732832670211792,
+      "learning_rate": 3.6e-05,
+      "loss": 0.032459765672683716,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.03299,
+      "step": 10,
+      "tokens/total": 303184,
+      "tokens/train_per_sec_per_gpu": 37.86,
+      "tokens/trainable": 4474
+    },
+    {
+      "epoch": 0.04296875,
+      "grad_norm": 1.0495871305465698,
+      "learning_rate": 4e-05,
+      "loss": 0.02137758582830429,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.02161,
+      "step": 11,
+      "tokens/total": 333296,
+      "tokens/train_per_sec_per_gpu": 33.24,
+      "tokens/trainable": 4942
+    },
+    {
+      "epoch": 0.046875,
+      "grad_norm": 2.064997673034668,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.04079515486955643,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.04164,
+      "step": 12,
+      "tokens/total": 363840,
+      "tokens/train_per_sec_per_gpu": 34.35,
+      "tokens/trainable": 5424
+    },
+    {
+      "epoch": 0.05078125,
+      "grad_norm": 2.9026453495025635,
+      "learning_rate": 4.8e-05,
+      "loss": 0.06431213766336441,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.06643,
+      "step": 13,
+      "tokens/total": 394112,
+      "tokens/train_per_sec_per_gpu": 35.34,
+      "tokens/trainable": 5855
+    },
+    {
+      "epoch": 0.0546875,
+      "grad_norm": 2.4172568321228027,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.07758771628141403,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.08068,
+      "step": 14,
+      "tokens/total": 424656,
+      "tokens/train_per_sec_per_gpu": 34.9,
+      "tokens/trainable": 6344
+    },
+    {
+      "epoch": 0.05859375,
+      "grad_norm": 2.6533877849578857,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 0.0820830836892128,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.08555,
+      "step": 15,
+      "tokens/total": 455152,
+      "tokens/train_per_sec_per_gpu": 36.2,
+      "tokens/trainable": 6825
+    },
+    {
+      "epoch": 0.0625,
+      "grad_norm": 2.7390506267547607,
+      "learning_rate": 6e-05,
+      "loss": 0.053628768771886826,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.05509,
+      "step": 16,
+      "tokens/total": 485328,
+      "tokens/train_per_sec_per_gpu": 30.12,
+      "tokens/trainable": 7230
+    },
+    {
+      "epoch": 0.06640625,
+      "grad_norm": 0.8172504305839539,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.016706150025129318,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01685,
+      "step": 17,
+      "tokens/total": 515744,
+      "tokens/train_per_sec_per_gpu": 36.07,
+      "tokens/trainable": 7699
+    },
+    {
+      "epoch": 0.0703125,
+      "grad_norm": 0.7143669724464417,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.021398911252617836,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02163,
+      "step": 18,
+      "tokens/total": 546144,
+      "tokens/train_per_sec_per_gpu": 33.9,
+      "tokens/trainable": 8184
+    },
+    {
+      "epoch": 0.07421875,
+      "grad_norm": 2.224132537841797,
+      "learning_rate": 7.2e-05,
+      "loss": 0.016565855592489243,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.0167,
+      "step": 19,
+      "tokens/total": 576560,
+      "tokens/train_per_sec_per_gpu": 35.01,
+      "tokens/trainable": 8675
+    },
+    {
+      "epoch": 0.078125,
+      "grad_norm": 1.9165221452713013,
+      "learning_rate": 7.6e-05,
+      "loss": 0.038831885904073715,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.0396,
+      "step": 20,
+      "tokens/total": 607056,
+      "tokens/train_per_sec_per_gpu": 39.89,
+      "tokens/trainable": 9189
+    },
+    {
+      "epoch": 0.08203125,
+      "grad_norm": 0.9385339617729187,
+      "learning_rate": 8e-05,
+      "loss": 0.02326519787311554,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02354,
+      "step": 21,
+      "tokens/total": 637488,
+      "tokens/train_per_sec_per_gpu": 31.74,
+      "tokens/trainable": 9614
+    },
+    {
+      "epoch": 0.0859375,
+      "grad_norm": 1.3244878053665161,
+      "learning_rate": 8.4e-05,
+      "loss": 0.04611007124185562,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.04719,
+      "step": 22,
+      "tokens/total": 667632,
+      "tokens/train_per_sec_per_gpu": 36.23,
+      "tokens/trainable": 10086
+    },
+    {
+      "epoch": 0.08984375,
+      "grad_norm": 1.130301594734192,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.03128095716238022,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.03178,
+      "step": 23,
+      "tokens/total": 697872,
+      "tokens/train_per_sec_per_gpu": 32.92,
+      "tokens/trainable": 10535
+    },
+    {
+      "epoch": 0.09375,
+      "grad_norm": 1.9159126281738281,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 0.042277175933122635,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.04318,
+      "step": 24,
+      "tokens/total": 728432,
+      "tokens/train_per_sec_per_gpu": 31.63,
+      "tokens/trainable": 10993
+    },
+    {
+      "epoch": 0.09765625,
+      "grad_norm": 0.5571489334106445,
+      "learning_rate": 9.6e-05,
+      "loss": 0.025266434997320175,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02559,
+      "step": 25,
+      "tokens/total": 756656,
+      "tokens/train_per_sec_per_gpu": 40.08,
+      "tokens/trainable": 11417
+    },
+    {
+      "epoch": 0.1015625,
+      "grad_norm": 0.8746730089187622,
+      "learning_rate": 0.0001,
+      "loss": 0.022446369752287865,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.0227,
+      "step": 26,
+      "tokens/total": 787120,
+      "tokens/train_per_sec_per_gpu": 34.39,
+      "tokens/trainable": 11874
+    },
+    {
+      "epoch": 0.10546875,
+      "grad_norm": 1.505344271659851,
+      "learning_rate": 9.99990636831587e-05,
+      "loss": 0.02043886110186577,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02065,
+      "step": 27,
+      "tokens/total": 817504,
+      "tokens/train_per_sec_per_gpu": 31.74,
+      "tokens/trainable": 12316
+    },
+    {
+      "epoch": 0.109375,
+      "grad_norm": 1.0260790586471558,
+      "learning_rate": 9.999625477159879e-05,
+      "loss": 0.021617872640490532,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02185,
+      "step": 28,
+      "tokens/total": 848128,
+      "tokens/train_per_sec_per_gpu": 37.86,
+      "tokens/trainable": 12800
+    },
+    {
+      "epoch": 0.11328125,
+      "grad_norm": 0.8115366101264954,
+      "learning_rate": 9.999157338221051e-05,
+      "loss": 0.020092125982046127,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0203,
+      "step": 29,
+      "tokens/total": 878448,
+      "tokens/train_per_sec_per_gpu": 35.57,
+      "tokens/trainable": 13270
+    },
+    {
+      "epoch": 0.1171875,
+      "grad_norm": 0.5937158465385437,
+      "learning_rate": 9.998501970980562e-05,
+      "loss": 0.014808757230639458,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.01492,
+      "step": 30,
+      "tokens/total": 909008,
+      "tokens/train_per_sec_per_gpu": 35.43,
+      "tokens/trainable": 13752
+    },
+    {
+      "epoch": 0.12109375,
+      "grad_norm": 0.7298683524131775,
+      "learning_rate": 9.997659402710915e-05,
+      "loss": 0.019437074661254883,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01963,
+      "step": 31,
+      "tokens/total": 939312,
+      "tokens/train_per_sec_per_gpu": 34.12,
+      "tokens/trainable": 14218
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 0.17546068131923676,
+      "learning_rate": 9.996629668474818e-05,
+      "loss": 0.003190835705026984,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.0032,
+      "step": 32,
+      "tokens/total": 969264,
+      "tokens/train_per_sec_per_gpu": 34.52,
+      "tokens/trainable": 14655
+    },
+    {
+      "epoch": 0.12890625,
+      "grad_norm": 0.8119296431541443,
+      "learning_rate": 9.995412811123711e-05,
+      "loss": 0.021450746804475784,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.02168,
+      "step": 33,
+      "tokens/total": 999792,
+      "tokens/train_per_sec_per_gpu": 33.42,
+      "tokens/trainable": 15087
+    },
+    {
+      "epoch": 0.1328125,
+      "grad_norm": 0.7597190141677856,
+      "learning_rate": 9.994008881295999e-05,
+      "loss": 0.02241288125514984,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02267,
+      "step": 34,
+      "tokens/total": 1029840,
+      "tokens/train_per_sec_per_gpu": 34.0,
+      "tokens/trainable": 15548
+    },
+    {
+      "epoch": 0.13671875,
+      "grad_norm": 2.3171119689941406,
+      "learning_rate": 9.992417937414932e-05,
+      "loss": 0.021416299045085907,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.02165,
+      "step": 35,
+      "tokens/total": 1060416,
+      "tokens/train_per_sec_per_gpu": 33.79,
+      "tokens/trainable": 16037
+    },
+    {
+      "epoch": 0.140625,
+      "grad_norm": 0.4047625958919525,
+      "learning_rate": 9.99064004568618e-05,
+      "loss": 0.005595909897238016,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.72,
+      "memory/max_allocated (GiB)": 33.72,
+      "ppl": 1.00561,
+      "step": 36,
+      "tokens/total": 1090464,
+      "tokens/train_per_sec_per_gpu": 31.13,
+      "tokens/trainable": 16475
+    },
+    {
+      "epoch": 0.14453125,
+      "grad_norm": 0.5865066051483154,
+      "learning_rate": 9.988675280095074e-05,
+      "loss": 0.010679518803954124,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.01074,
+      "step": 37,
+      "tokens/total": 1120944,
+      "tokens/train_per_sec_per_gpu": 34.44,
+      "tokens/trainable": 16919
+    },
+    {
+      "epoch": 0.1484375,
+      "grad_norm": 1.583072304725647,
+      "learning_rate": 9.986523722403528e-05,
+      "loss": 0.02127654477953911,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0215,
+      "step": 38,
+      "tokens/total": 1151360,
+      "tokens/train_per_sec_per_gpu": 29.02,
+      "tokens/trainable": 17333
+    },
+    {
+      "epoch": 0.15234375,
+      "grad_norm": 1.30342698097229,
+      "learning_rate": 9.984185462146642e-05,
+      "loss": 0.025776617228984833,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.02611,
+      "step": 39,
+      "tokens/total": 1181728,
+      "tokens/train_per_sec_per_gpu": 36.2,
+      "tokens/trainable": 17806
+    },
+    {
+      "epoch": 0.15625,
+      "grad_norm": 1.1053258180618286,
+      "learning_rate": 9.98166059662897e-05,
+      "loss": 0.026606226339936256,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02696,
+      "step": 40,
+      "tokens/total": 1212064,
+      "tokens/train_per_sec_per_gpu": 35.67,
+      "tokens/trainable": 18290
+    },
+    {
+      "epoch": 0.16015625,
+      "grad_norm": 0.6078025102615356,
+      "learning_rate": 9.978949230920472e-05,
+      "loss": 0.017936185002326965,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.0181,
+      "step": 41,
+      "tokens/total": 1242592,
+      "tokens/train_per_sec_per_gpu": 32.05,
+      "tokens/trainable": 18747
+    },
+    {
+      "epoch": 0.1640625,
+      "grad_norm": 0.8376394510269165,
+      "learning_rate": 9.976051477852141e-05,
+      "loss": 0.02379559725522995,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02408,
+      "step": 42,
+      "tokens/total": 1272800,
+      "tokens/train_per_sec_per_gpu": 31.9,
+      "tokens/trainable": 19172
+    },
+    {
+      "epoch": 0.16796875,
+      "grad_norm": 0.4673662781715393,
+      "learning_rate": 9.972967458011312e-05,
+      "loss": 0.012735492549836636,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 34.04,
+      "memory/max_allocated (GiB)": 34.04,
+      "ppl": 1.01282,
+      "step": 43,
+      "tokens/total": 1303600,
+      "tokens/train_per_sec_per_gpu": 31.35,
+      "tokens/trainable": 19622
+    },
+    {
+      "epoch": 0.171875,
+      "grad_norm": 0.38821765780448914,
+      "learning_rate": 9.96969729973664e-05,
+      "loss": 0.008668653666973114,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00871,
+      "step": 44,
+      "tokens/total": 1333936,
+      "tokens/train_per_sec_per_gpu": 33.34,
+      "tokens/trainable": 20073
+    },
+    {
+      "epoch": 0.17578125,
+      "grad_norm": 0.7656640410423279,
+      "learning_rate": 9.966241139112754e-05,
+      "loss": 0.020857544615864754,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.02108,
+      "step": 45,
+      "tokens/total": 1364288,
+      "tokens/train_per_sec_per_gpu": 31.55,
+      "tokens/trainable": 20497
+    },
+    {
+      "epoch": 0.1796875,
+      "grad_norm": 0.6551951766014099,
+      "learning_rate": 9.96259911996461e-05,
+      "loss": 0.020156342536211014,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02036,
+      "step": 46,
+      "tokens/total": 1394368,
+      "tokens/train_per_sec_per_gpu": 34.35,
+      "tokens/trainable": 20940
+    },
+    {
+      "epoch": 0.18359375,
+      "grad_norm": 1.2237868309020996,
+      "learning_rate": 9.958771393851491e-05,
+      "loss": 0.015184836462140083,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.98,
+      "memory/max_allocated (GiB)": 33.98,
+      "ppl": 1.0153,
+      "step": 47,
+      "tokens/total": 1425024,
+      "tokens/train_per_sec_per_gpu": 35.07,
+      "tokens/trainable": 21405
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 1.571843147277832,
+      "learning_rate": 9.954758120060702e-05,
+      "loss": 0.01640632003545761,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01654,
+      "step": 48,
+      "tokens/total": 1455392,
+      "tokens/train_per_sec_per_gpu": 35.15,
+      "tokens/trainable": 21828
+    },
+    {
+      "epoch": 0.19140625,
+      "grad_norm": 1.3066037893295288,
+      "learning_rate": 9.950559465600948e-05,
+      "loss": 0.017745690420269966,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.0179,
+      "step": 49,
+      "tokens/total": 1485616,
+      "tokens/train_per_sec_per_gpu": 29.54,
+      "tokens/trainable": 22241
+    },
+    {
+      "epoch": 0.1953125,
+      "grad_norm": 0.6320858597755432,
+      "learning_rate": 9.946175605195379e-05,
+      "loss": 0.009900711476802826,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00995,
+      "step": 50,
+      "tokens/total": 1515984,
+      "tokens/train_per_sec_per_gpu": 36.38,
+      "tokens/trainable": 22688
+    },
+    {
+      "epoch": 0.19921875,
+      "grad_norm": 0.3268744647502899,
+      "learning_rate": 9.941606721274322e-05,
+      "loss": 0.006143633276224136,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00616,
+      "step": 51,
+      "tokens/total": 1546224,
+      "tokens/train_per_sec_per_gpu": 32.16,
+      "tokens/trainable": 23136
+    },
+    {
+      "epoch": 0.203125,
+      "grad_norm": 0.3406257629394531,
+      "learning_rate": 9.936853003967685e-05,
+      "loss": 0.004097095225006342,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00411,
+      "step": 52,
+      "tokens/total": 1576528,
+      "tokens/train_per_sec_per_gpu": 37.04,
+      "tokens/trainable": 23631
+    },
+    {
+      "epoch": 0.20703125,
+      "grad_norm": 1.1476560831069946,
+      "learning_rate": 9.93191465109705e-05,
+      "loss": 0.028221281245350838,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02862,
+      "step": 53,
+      "tokens/total": 1606768,
+      "tokens/train_per_sec_per_gpu": 34.12,
+      "tokens/trainable": 24096
+    },
+    {
+      "epoch": 0.2109375,
+      "grad_norm": 0.451351135969162,
+      "learning_rate": 9.926791868167438e-05,
+      "loss": 0.0078392019495368,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.66,
+      "memory/max_allocated (GiB)": 33.66,
+      "ppl": 1.00787,
+      "step": 54,
+      "tokens/total": 1636640,
+      "tokens/train_per_sec_per_gpu": 37.82,
+      "tokens/trainable": 24569
+    },
+    {
+      "epoch": 0.21484375,
+      "grad_norm": 0.8821313977241516,
+      "learning_rate": 9.921484868358753e-05,
+      "loss": 0.017667783424258232,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01782,
+      "step": 55,
+      "tokens/total": 1667136,
+      "tokens/train_per_sec_per_gpu": 34.02,
+      "tokens/trainable": 25067
+    },
+    {
+      "epoch": 0.21875,
+      "grad_norm": 0.6063156723976135,
+      "learning_rate": 9.915993872516924e-05,
+      "loss": 0.012062735855579376,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.01214,
+      "step": 56,
+      "tokens/total": 1697472,
+      "tokens/train_per_sec_per_gpu": 36.6,
+      "tokens/trainable": 25541
+    },
+    {
+      "epoch": 0.22265625,
+      "grad_norm": 0.7402717471122742,
+      "learning_rate": 9.9103191091447e-05,
+      "loss": 0.030087217688560486,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03054,
+      "step": 57,
+      "tokens/total": 1727680,
+      "tokens/train_per_sec_per_gpu": 33.28,
+      "tokens/trainable": 25965
+    },
+    {
+      "epoch": 0.2265625,
+      "grad_norm": 0.5932730436325073,
+      "learning_rate": 9.904460814392147e-05,
+      "loss": 0.013998140580952168,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0141,
+      "step": 58,
+      "tokens/total": 1758192,
+      "tokens/train_per_sec_per_gpu": 33.94,
+      "tokens/trainable": 26409
+    },
+    {
+      "epoch": 0.23046875,
+      "grad_norm": 0.6833471059799194,
+      "learning_rate": 9.898419232046825e-05,
+      "loss": 0.009387579746544361,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00943,
+      "step": 59,
+      "tokens/total": 1788816,
+      "tokens/train_per_sec_per_gpu": 34.07,
+      "tokens/trainable": 26883
+    },
+    {
+      "epoch": 0.234375,
+      "grad_norm": 0.49946627020835876,
+      "learning_rate": 9.892194613523633e-05,
+      "loss": 0.014185642823576927,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01429,
+      "step": 60,
+      "tokens/total": 1819056,
+      "tokens/train_per_sec_per_gpu": 33.72,
+      "tokens/trainable": 27323
+    },
+    {
+      "epoch": 0.23828125,
+      "grad_norm": 0.17088162899017334,
+      "learning_rate": 9.885787217854357e-05,
+      "loss": 0.003224549815058708,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00323,
+      "step": 61,
+      "tokens/total": 1849664,
+      "tokens/train_per_sec_per_gpu": 39.22,
+      "tokens/trainable": 27830
+    },
+    {
+      "epoch": 0.2421875,
+      "grad_norm": 0.5056981444358826,
+      "learning_rate": 9.879197311676887e-05,
+      "loss": 0.012983381748199463,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 34.02,
+      "memory/max_allocated (GiB)": 34.02,
+      "ppl": 1.01307,
+      "step": 62,
+      "tokens/total": 1880272,
+      "tokens/train_per_sec_per_gpu": 32.36,
+      "tokens/trainable": 28281
+    },
+    {
+      "epoch": 0.24609375,
+      "grad_norm": 2.0115866661071777,
+      "learning_rate": 9.872425169224113e-05,
+      "loss": 0.02842879481613636,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.02884,
+      "step": 63,
+      "tokens/total": 1910752,
+      "tokens/train_per_sec_per_gpu": 32.32,
+      "tokens/trainable": 28742
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8811206221580505,
+      "learning_rate": 9.865471072312528e-05,
+      "loss": 0.016607588157057762,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01675,
+      "step": 64,
+      "tokens/total": 1940848,
+      "tokens/train_per_sec_per_gpu": 33.58,
+      "tokens/trainable": 29183
+    },
+    {
+      "epoch": 0.25390625,
+      "grad_norm": 0.12346348166465759,
+      "learning_rate": 9.858335310330492e-05,
+      "loss": 0.003063689451664686,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00307,
+      "step": 65,
+      "tokens/total": 1971344,
+      "tokens/train_per_sec_per_gpu": 31.61,
+      "tokens/trainable": 29648
+    },
+    {
+      "epoch": 0.2578125,
+      "grad_norm": 0.28303098678588867,
+      "learning_rate": 9.851018180226185e-05,
+      "loss": 0.00448613939806819,
+      "memory/device_reserved (GiB)": 35.06,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.0045,
+      "step": 66,
+      "tokens/total": 2001712,
+      "tokens/train_per_sec_per_gpu": 33.54,
+      "tokens/trainable": 30075
+    },
+    {
+      "epoch": 0.26171875,
+      "grad_norm": 0.7989771962165833,
+      "learning_rate": 9.843519986495259e-05,
+      "loss": 0.008180832490324974,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.00821,
+      "step": 67,
+      "tokens/total": 2029936,
+      "tokens/train_per_sec_per_gpu": 37.66,
+      "tokens/trainable": 30546
+    },
+    {
+      "epoch": 0.265625,
+      "grad_norm": 0.12454555928707123,
+      "learning_rate": 9.835841041168162e-05,
+      "loss": 0.0016878236783668399,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00169,
+      "step": 68,
+      "tokens/total": 2060384,
+      "tokens/train_per_sec_per_gpu": 35.16,
+      "tokens/trainable": 31025
+    },
+    {
+      "epoch": 0.26953125,
+      "grad_norm": 1.7390296459197998,
+      "learning_rate": 9.82798166379715e-05,
+      "loss": 0.016503486782312393,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.01664,
+      "step": 69,
+      "tokens/total": 2090688,
+      "tokens/train_per_sec_per_gpu": 33.9,
+      "tokens/trainable": 31505
+    },
+    {
+      "epoch": 0.2734375,
+      "grad_norm": 0.968937337398529,
+      "learning_rate": 9.819942181443002e-05,
+      "loss": 0.01685403287410736,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.017,
+      "step": 70,
+      "tokens/total": 2120848,
+      "tokens/train_per_sec_per_gpu": 32.32,
+      "tokens/trainable": 31956
+    },
+    {
+      "epoch": 0.27734375,
+      "grad_norm": 0.3503779470920563,
+      "learning_rate": 9.811722928661392e-05,
+      "loss": 0.005242045037448406,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.35,
+      "memory/max_allocated (GiB)": 33.35,
+      "ppl": 1.00526,
+      "step": 71,
+      "tokens/total": 2149120,
+      "tokens/train_per_sec_per_gpu": 33.45,
+      "tokens/trainable": 32374
+    },
+    {
+      "epoch": 0.28125,
+      "grad_norm": 0.5608110427856445,
+      "learning_rate": 9.803324247488975e-05,
+      "loss": 0.013865578919649124,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01396,
+      "step": 72,
+      "tokens/total": 2179472,
+      "tokens/train_per_sec_per_gpu": 34.56,
+      "tokens/trainable": 32833
+    },
+    {
+      "epoch": 0.28515625,
+      "grad_norm": 1.2297694683074951,
+      "learning_rate": 9.794746487429161e-05,
+      "loss": 0.017295369878411293,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01745,
+      "step": 73,
+      "tokens/total": 2209984,
+      "tokens/train_per_sec_per_gpu": 35.72,
+      "tokens/trainable": 33304
+    },
+    {
+      "epoch": 0.2890625,
+      "grad_norm": 0.1584348827600479,
+      "learning_rate": 9.785990005437554e-05,
+      "loss": 0.004732954781502485,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00474,
+      "step": 74,
+      "tokens/total": 2240512,
+      "tokens/train_per_sec_per_gpu": 32.44,
+      "tokens/trainable": 33771
+    },
+    {
+      "epoch": 0.29296875,
+      "grad_norm": 0.3246576488018036,
+      "learning_rate": 9.777055165907117e-05,
+      "loss": 0.013291046023368835,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01338,
+      "step": 75,
+      "tokens/total": 2270608,
+      "tokens/train_per_sec_per_gpu": 34.64,
+      "tokens/trainable": 34186
+    },
+    {
+      "epoch": 0.296875,
+      "grad_norm": 1.1803282499313354,
+      "learning_rate": 9.767942340652993e-05,
+      "loss": 0.014760692603886127,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01487,
+      "step": 76,
+      "tokens/total": 2300864,
+      "tokens/train_per_sec_per_gpu": 34.96,
+      "tokens/trainable": 34667
+    },
+    {
+      "epoch": 0.30078125,
+      "grad_norm": 2.001532793045044,
+      "learning_rate": 9.758651908897035e-05,
+      "loss": 0.06761633604764938,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.76,
+      "memory/max_allocated (GiB)": 33.76,
+      "ppl": 1.06995,
+      "step": 77,
+      "tokens/total": 2330960,
+      "tokens/train_per_sec_per_gpu": 31.81,
+      "tokens/trainable": 35078
+    },
+    {
+      "epoch": 0.3046875,
+      "grad_norm": 0.29207468032836914,
+      "learning_rate": 9.749184257252033e-05,
+      "loss": 0.006066338159143925,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.00608,
+      "step": 78,
+      "tokens/total": 2361584,
+      "tokens/train_per_sec_per_gpu": 34.95,
+      "tokens/trainable": 35550
+    },
+    {
+      "epoch": 0.30859375,
+      "grad_norm": 0.5561455488204956,
+      "learning_rate": 9.739539779705614e-05,
+      "loss": 0.033555109053850174,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.03412,
+      "step": 79,
+      "tokens/total": 2391936,
+      "tokens/train_per_sec_per_gpu": 29.55,
+      "tokens/trainable": 35977
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 0.4088423550128937,
+      "learning_rate": 9.729718877603861e-05,
+      "loss": 0.014240232296288013,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01434,
+      "step": 80,
+      "tokens/total": 2422432,
+      "tokens/train_per_sec_per_gpu": 38.29,
+      "tokens/trainable": 36462
+    },
+    {
+      "epoch": 0.31640625,
+      "grad_norm": 0.2712189853191376,
+      "learning_rate": 9.719721959634592e-05,
+      "loss": 0.011144062504172325,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01121,
+      "step": 81,
+      "tokens/total": 2452752,
+      "tokens/train_per_sec_per_gpu": 33.24,
+      "tokens/trainable": 36919
+    },
+    {
+      "epoch": 0.3203125,
+      "grad_norm": 0.40229442715644836,
+      "learning_rate": 9.709549441810375e-05,
+      "loss": 0.012572492472827435,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.44,
+      "memory/max_allocated (GiB)": 33.44,
+      "ppl": 1.01265,
+      "step": 82,
+      "tokens/total": 2481216,
+      "tokens/train_per_sec_per_gpu": 33.48,
+      "tokens/trainable": 37379
+    },
+    {
+      "epoch": 0.32421875,
+      "grad_norm": 3.2695465087890625,
+      "learning_rate": 9.699201747451195e-05,
+      "loss": 0.012744108214974403,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01283,
+      "step": 83,
+      "tokens/total": 2511488,
+      "tokens/train_per_sec_per_gpu": 33.12,
+      "tokens/trainable": 37832
+    },
+    {
+      "epoch": 0.328125,
+      "grad_norm": 0.42963695526123047,
+      "learning_rate": 9.688679307166854e-05,
+      "loss": 0.014645008370280266,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01475,
+      "step": 84,
+      "tokens/total": 2541984,
+      "tokens/train_per_sec_per_gpu": 34.0,
+      "tokens/trainable": 38303
+    },
+    {
+      "epoch": 0.33203125,
+      "grad_norm": 0.6282569169998169,
+      "learning_rate": 9.677982558839042e-05,
+      "loss": 0.009028157219290733,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.69,
+      "memory/max_allocated (GiB)": 33.69,
+      "ppl": 1.00907,
+      "step": 85,
+      "tokens/total": 2572224,
+      "tokens/train_per_sec_per_gpu": 33.71,
+      "tokens/trainable": 38758
+    },
+    {
+      "epoch": 0.3359375,
+      "grad_norm": 0.5804964900016785,
+      "learning_rate": 9.66711194760312e-05,
+      "loss": 0.010930215008556843,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01099,
+      "step": 86,
+      "tokens/total": 2602704,
+      "tokens/train_per_sec_per_gpu": 30.71,
+      "tokens/trainable": 39179
+    },
+    {
+      "epoch": 0.33984375,
+      "grad_norm": 0.3565916121006012,
+      "learning_rate": 9.656067925829593e-05,
+      "loss": 0.011934653855860233,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01201,
+      "step": 87,
+      "tokens/total": 2633072,
+      "tokens/train_per_sec_per_gpu": 35.39,
+      "tokens/trainable": 39679
+    },
+    {
+      "epoch": 0.34375,
+      "grad_norm": 0.5799129009246826,
+      "learning_rate": 9.644850953105288e-05,
+      "loss": 0.010369446128606796,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.01042,
+      "step": 88,
+      "tokens/total": 2663232,
+      "tokens/train_per_sec_per_gpu": 32.76,
+      "tokens/trainable": 40118
+    },
+    {
+      "epoch": 0.34765625,
+      "grad_norm": 0.4171448051929474,
+      "learning_rate": 9.633461496214225e-05,
+      "loss": 0.012263461016118526,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.01234,
+      "step": 89,
+      "tokens/total": 2693696,
+      "tokens/train_per_sec_per_gpu": 32.95,
+      "tokens/trainable": 40571
+    },
+    {
+      "epoch": 0.3515625,
+      "grad_norm": 0.11327923834323883,
+      "learning_rate": 9.621900029118195e-05,
+      "loss": 0.0015271489974111319,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00153,
+      "step": 90,
+      "tokens/total": 2723984,
+      "tokens/train_per_sec_per_gpu": 30.97,
+      "tokens/trainable": 40997
+    },
+    {
+      "epoch": 0.35546875,
+      "grad_norm": 0.49398481845855713,
+      "learning_rate": 9.610167032937036e-05,
+      "loss": 0.0075433989986777306,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00757,
+      "step": 91,
+      "tokens/total": 2754240,
+      "tokens/train_per_sec_per_gpu": 36.54,
+      "tokens/trainable": 41462
+    },
+    {
+      "epoch": 0.359375,
+      "grad_norm": 0.3695297837257385,
+      "learning_rate": 9.598262995928611e-05,
+      "loss": 0.0067325979471206665,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00676,
+      "step": 92,
+      "tokens/total": 2784672,
+      "tokens/train_per_sec_per_gpu": 36.93,
+      "tokens/trainable": 41927
+    },
+    {
+      "epoch": 0.36328125,
+      "grad_norm": 0.6571913957595825,
+      "learning_rate": 9.586188413468492e-05,
+      "loss": 0.010379074141383171,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01043,
+      "step": 93,
+      "tokens/total": 2815120,
+      "tokens/train_per_sec_per_gpu": 34.92,
+      "tokens/trainable": 42397
+    },
+    {
+      "epoch": 0.3671875,
+      "grad_norm": 0.09913541376590729,
+      "learning_rate": 9.57394378802934e-05,
+      "loss": 0.0005878547672182322,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.00059,
+      "step": 94,
+      "tokens/total": 2845600,
+      "tokens/train_per_sec_per_gpu": 36.69,
+      "tokens/trainable": 42883
+    },
+    {
+      "epoch": 0.37109375,
+      "grad_norm": 0.3970830738544464,
+      "learning_rate": 9.56152962916e-05,
+      "loss": 0.005916361231356859,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00593,
+      "step": 95,
+      "tokens/total": 2875952,
+      "tokens/train_per_sec_per_gpu": 38.83,
+      "tokens/trainable": 43381
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 0.1583622545003891,
+      "learning_rate": 9.548946453464296e-05,
+      "loss": 0.0020841225050389767,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00209,
+      "step": 96,
+      "tokens/total": 2906288,
+      "tokens/train_per_sec_per_gpu": 33.1,
+      "tokens/trainable": 43824
+    },
+    {
+      "epoch": 0.37890625,
+      "grad_norm": 0.42756563425064087,
+      "learning_rate": 9.53619478457953e-05,
+      "loss": 0.004547779448330402,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00456,
+      "step": 97,
+      "tokens/total": 2936752,
+      "tokens/train_per_sec_per_gpu": 31.67,
+      "tokens/trainable": 44255
+    },
+    {
+      "epoch": 0.3828125,
+      "grad_norm": 0.9736410975456238,
+      "learning_rate": 9.523275153154695e-05,
+      "loss": 0.04175024852156639,
+      "memory/device_reserved (GiB)": 35.19,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.04263,
+      "step": 98,
+      "tokens/total": 2967072,
+      "tokens/train_per_sec_per_gpu": 37.88,
+      "tokens/trainable": 44739
+    },
+    {
+      "epoch": 0.38671875,
+      "grad_norm": 0.8381142020225525,
+      "learning_rate": 9.51018809682839e-05,
+      "loss": 0.013025594875216484,
+      "memory/device_reserved (GiB)": 35.76,
+      "memory/max_active (GiB)": 34.03,
+      "memory/max_allocated (GiB)": 34.03,
+      "ppl": 1.01311,
+      "step": 99,
+      "tokens/total": 2997664,
+      "tokens/train_per_sec_per_gpu": 36.31,
+      "tokens/trainable": 45202
+    },
+    {
+      "epoch": 0.390625,
+      "grad_norm": 0.22207525372505188,
+      "learning_rate": 9.49693416020645e-05,
+      "loss": 0.004689273424446583,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.0047,
+      "step": 100,
+      "tokens/total": 3028032,
+      "tokens/train_per_sec_per_gpu": 33.14,
+      "tokens/trainable": 45646
+    },
+    {
+      "epoch": 0.39453125,
+      "grad_norm": 0.28477466106414795,
+      "learning_rate": 9.483513894839276e-05,
+      "loss": 0.010324095375835896,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01038,
+      "step": 101,
+      "tokens/total": 3058272,
+      "tokens/train_per_sec_per_gpu": 33.09,
+      "tokens/trainable": 46089
+    },
+    {
+      "epoch": 0.3984375,
+      "grad_norm": 0.5098783373832703,
+      "learning_rate": 9.469927859198888e-05,
+      "loss": 0.028538472950458527,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.02895,
+      "step": 102,
+      "tokens/total": 3088848,
+      "tokens/train_per_sec_per_gpu": 32.63,
+      "tokens/trainable": 46547
+    },
+    {
+      "epoch": 0.40234375,
+      "grad_norm": 0.3906193673610687,
+      "learning_rate": 9.456176618655689e-05,
+      "loss": 0.017888592556118965,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.01805,
+      "step": 103,
+      "tokens/total": 3119424,
+      "tokens/train_per_sec_per_gpu": 32.02,
+      "tokens/trainable": 47014
+    },
+    {
+      "epoch": 0.40625,
+      "grad_norm": 0.5902650952339172,
+      "learning_rate": 9.442260745454927e-05,
+      "loss": 0.03132036700844765,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.03182,
+      "step": 104,
+      "tokens/total": 3149696,
+      "tokens/train_per_sec_per_gpu": 37.8,
+      "tokens/trainable": 47503
+    },
+    {
+      "epoch": 0.41015625,
+      "grad_norm": 0.2547222375869751,
+      "learning_rate": 9.428180818692884e-05,
+      "loss": 0.014553382061421871,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01466,
+      "step": 105,
+      "tokens/total": 3180064,
+      "tokens/train_per_sec_per_gpu": 36.37,
+      "tokens/trainable": 47996
+    },
+    {
+      "epoch": 0.4140625,
+      "grad_norm": 0.49491167068481445,
+      "learning_rate": 9.413937424292791e-05,
+      "loss": 0.0295359306037426,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.02998,
+      "step": 106,
+      "tokens/total": 3208320,
+      "tokens/train_per_sec_per_gpu": 34.18,
+      "tokens/trainable": 48447
+    },
+    {
+      "epoch": 0.41796875,
+      "grad_norm": 0.481608510017395,
+      "learning_rate": 9.399531154980424e-05,
+      "loss": 0.017734361812472343,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01789,
+      "step": 107,
+      "tokens/total": 3238416,
+      "tokens/train_per_sec_per_gpu": 30.78,
+      "tokens/trainable": 48889
+    },
+    {
+      "epoch": 0.421875,
+      "grad_norm": 0.3913033604621887,
+      "learning_rate": 9.384962610259455e-05,
+      "loss": 0.022145511582493782,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02239,
+      "step": 108,
+      "tokens/total": 3268784,
+      "tokens/train_per_sec_per_gpu": 38.03,
+      "tokens/trainable": 49360
+    },
+    {
+      "epoch": 0.42578125,
+      "grad_norm": 0.1751519739627838,
+      "learning_rate": 9.370232396386494e-05,
+      "loss": 0.0074960459023714066,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00752,
+      "step": 109,
+      "tokens/total": 3298912,
+      "tokens/train_per_sec_per_gpu": 36.89,
+      "tokens/trainable": 49827
+    },
+    {
+      "epoch": 0.4296875,
+      "grad_norm": 0.28872016072273254,
+      "learning_rate": 9.355341126345868e-05,
+      "loss": 0.01847866177558899,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.01865,
+      "step": 110,
+      "tokens/total": 3329232,
+      "tokens/train_per_sec_per_gpu": 34.36,
+      "tokens/trainable": 50267
+    },
+    {
+      "epoch": 0.43359375,
+      "grad_norm": 0.3326593041419983,
+      "learning_rate": 9.340289419824107e-05,
+      "loss": 0.013967925682663918,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.01407,
+      "step": 111,
+      "tokens/total": 3359440,
+      "tokens/train_per_sec_per_gpu": 33.47,
+      "tokens/trainable": 50711
+    },
+    {
+      "epoch": 0.4375,
+      "grad_norm": 0.47660934925079346,
+      "learning_rate": 9.325077903184159e-05,
+      "loss": 0.02260858193039894,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.02287,
+      "step": 112,
+      "tokens/total": 3389472,
+      "tokens/train_per_sec_per_gpu": 31.72,
+      "tokens/trainable": 51157
+    },
+    {
+      "epoch": 0.44140625,
+      "grad_norm": 0.3481523096561432,
+      "learning_rate": 9.30970720943932e-05,
+      "loss": 0.008511725813150406,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00855,
+      "step": 113,
+      "tokens/total": 3419920,
+      "tokens/train_per_sec_per_gpu": 32.79,
+      "tokens/trainable": 51613
+    },
+    {
+      "epoch": 0.4453125,
+      "grad_norm": 0.4288201928138733,
+      "learning_rate": 9.2941779782269e-05,
+      "loss": 0.02450375072658062,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.02481,
+      "step": 114,
+      "tokens/total": 3450176,
+      "tokens/train_per_sec_per_gpu": 29.19,
+      "tokens/trainable": 52017
+    },
+    {
+      "epoch": 0.44921875,
+      "grad_norm": 0.17284883558750153,
+      "learning_rate": 9.278490855781596e-05,
+      "loss": 0.007882107980549335,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00791,
+      "step": 115,
+      "tokens/total": 3480544,
+      "tokens/train_per_sec_per_gpu": 36.36,
+      "tokens/trainable": 52506
+    },
+    {
+      "epoch": 0.453125,
+      "grad_norm": 0.2693459689617157,
+      "learning_rate": 9.262646494908604e-05,
+      "loss": 0.005466018337756395,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00548,
+      "step": 116,
+      "tokens/total": 3510848,
+      "tokens/train_per_sec_per_gpu": 38.77,
+      "tokens/trainable": 53008
+    },
+    {
+      "epoch": 0.45703125,
+      "grad_norm": 0.48690056800842285,
+      "learning_rate": 9.246645554956457e-05,
+      "loss": 0.004998547025024891,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.00501,
+      "step": 117,
+      "tokens/total": 3541296,
+      "tokens/train_per_sec_per_gpu": 35.47,
+      "tokens/trainable": 53467
+    },
+    {
+      "epoch": 0.4609375,
+      "grad_norm": 0.43558549880981445,
+      "learning_rate": 9.230488701789578e-05,
+      "loss": 0.017219291999936104,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01737,
+      "step": 118,
+      "tokens/total": 3571936,
+      "tokens/train_per_sec_per_gpu": 32.96,
+      "tokens/trainable": 53923
+    },
+    {
+      "epoch": 0.46484375,
+      "grad_norm": 0.28566548228263855,
+      "learning_rate": 9.214176607760577e-05,
+      "loss": 0.00544874370098114,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00546,
+      "step": 119,
+      "tokens/total": 3602000,
+      "tokens/train_per_sec_per_gpu": 35.72,
+      "tokens/trainable": 54363
+    },
+    {
+      "epoch": 0.46875,
+      "grad_norm": 0.07611701637506485,
+      "learning_rate": 9.197709951682268e-05,
+      "loss": 0.0014362478395923972,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00144,
+      "step": 120,
+      "tokens/total": 3632288,
+      "tokens/train_per_sec_per_gpu": 35.29,
+      "tokens/trainable": 54839
+    },
+    {
+      "epoch": 0.47265625,
+      "grad_norm": 0.13201506435871124,
+      "learning_rate": 9.181089418799428e-05,
+      "loss": 0.0024022101424634457,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00241,
+      "step": 121,
+      "tokens/total": 3662608,
+      "tokens/train_per_sec_per_gpu": 32.7,
+      "tokens/trainable": 55267
+    },
+    {
+      "epoch": 0.4765625,
+      "grad_norm": 0.40975725650787354,
+      "learning_rate": 9.164315700760271e-05,
+      "loss": 0.008171994239091873,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.00821,
+      "step": 122,
+      "tokens/total": 3692864,
+      "tokens/train_per_sec_per_gpu": 30.6,
+      "tokens/trainable": 55700
+    },
+    {
+      "epoch": 0.48046875,
+      "grad_norm": 0.15150536596775055,
+      "learning_rate": 9.147389495587671e-05,
+      "loss": 0.0036906469613313675,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.0037,
+      "step": 123,
+      "tokens/total": 3722864,
+      "tokens/train_per_sec_per_gpu": 33.48,
+      "tokens/trainable": 56144
+    },
+    {
+      "epoch": 0.484375,
+      "grad_norm": 0.514573335647583,
+      "learning_rate": 9.130311507650116e-05,
+      "loss": 0.01570757105946541,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01583,
+      "step": 124,
+      "tokens/total": 3753056,
+      "tokens/train_per_sec_per_gpu": 33.18,
+      "tokens/trainable": 56589
+    },
+    {
+      "epoch": 0.48828125,
+      "grad_norm": 0.43223315477371216,
+      "learning_rate": 9.113082447632394e-05,
+      "loss": 0.007733152247965336,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00776,
+      "step": 125,
+      "tokens/total": 3783312,
+      "tokens/train_per_sec_per_gpu": 36.78,
+      "tokens/trainable": 57081
+    },
+    {
+      "epoch": 0.4921875,
+      "grad_norm": 0.2972177565097809,
+      "learning_rate": 9.09570303250602e-05,
+      "loss": 0.008276824839413166,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00831,
+      "step": 126,
+      "tokens/total": 3813712,
+      "tokens/train_per_sec_per_gpu": 32.97,
+      "tokens/trainable": 57527
+    },
+    {
+      "epoch": 0.49609375,
+      "grad_norm": 0.6973468065261841,
+      "learning_rate": 9.078173985499394e-05,
+      "loss": 0.011385848745703697,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.01145,
+      "step": 127,
+      "tokens/total": 3844336,
+      "tokens/train_per_sec_per_gpu": 34.98,
+      "tokens/trainable": 58005
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.11440204083919525,
+      "learning_rate": 9.060496036067713e-05,
+      "loss": 0.0026649129576981068,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.00267,
+      "step": 128,
+      "tokens/total": 3872688,
+      "tokens/train_per_sec_per_gpu": 35.34,
+      "tokens/trainable": 58454
+    },
+    {
+      "epoch": 0.50390625,
+      "grad_norm": 0.4737952649593353,
+      "learning_rate": 9.042669919862615e-05,
+      "loss": 0.011923262849450111,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.64,
+      "memory/max_allocated (GiB)": 33.64,
+      "ppl": 1.01199,
+      "step": 129,
+      "tokens/total": 3902640,
+      "tokens/train_per_sec_per_gpu": 32.11,
+      "tokens/trainable": 58896
+    },
+    {
+      "epoch": 0.5078125,
+      "grad_norm": 0.22904613614082336,
+      "learning_rate": 9.024696378701557e-05,
+      "loss": 0.003364444011822343,
+      "memory/device_reserved (GiB)": 34.68,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00337,
+      "step": 130,
+      "tokens/total": 3933104,
+      "tokens/train_per_sec_per_gpu": 35.96,
+      "tokens/trainable": 59356
+    },
+    {
+      "epoch": 0.51171875,
+      "grad_norm": 0.6996368169784546,
+      "learning_rate": 9.006576160536948e-05,
+      "loss": 0.0145557951182127,
+      "memory/device_reserved (GiB)": 34.92,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.01466,
+      "step": 131,
+      "tokens/total": 3963184,
+      "tokens/train_per_sec_per_gpu": 32.33,
+      "tokens/trainable": 59792
+    },
+    {
+      "epoch": 0.515625,
+      "grad_norm": 0.18643876910209656,
+      "learning_rate": 8.988310019425035e-05,
+      "loss": 0.004168116021901369,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00418,
+      "step": 132,
+      "tokens/total": 3993648,
+      "tokens/train_per_sec_per_gpu": 35.79,
+      "tokens/trainable": 60266
+    },
+    {
+      "epoch": 0.51953125,
+      "grad_norm": 0.3754265606403351,
+      "learning_rate": 8.969898715494506e-05,
+      "loss": 0.01239765528589487,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01247,
+      "step": 133,
+      "tokens/total": 4024000,
+      "tokens/train_per_sec_per_gpu": 35.46,
+      "tokens/trainable": 60759
+    },
+    {
+      "epoch": 0.5234375,
+      "grad_norm": 0.5720171332359314,
+      "learning_rate": 8.951343014914869e-05,
+      "loss": 0.028643302619457245,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.02906,
+      "step": 134,
+      "tokens/total": 4052480,
+      "tokens/train_per_sec_per_gpu": 31.88,
+      "tokens/trainable": 61204
+    },
+    {
+      "epoch": 0.52734375,
+      "grad_norm": 0.21384330093860626,
+      "learning_rate": 8.932643689864568e-05,
+      "loss": 0.005144410766661167,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00516,
+      "step": 135,
+      "tokens/total": 4082864,
+      "tokens/train_per_sec_per_gpu": 36.1,
+      "tokens/trainable": 61684
+    },
+    {
+      "epoch": 0.53125,
+      "grad_norm": 0.3025348484516144,
+      "learning_rate": 8.913801518498845e-05,
+      "loss": 0.011437858454883099,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.7,
+      "memory/max_allocated (GiB)": 33.7,
+      "ppl": 1.0115,
+      "step": 136,
+      "tokens/total": 4113040,
+      "tokens/train_per_sec_per_gpu": 38.36,
+      "tokens/trainable": 62159
+    },
+    {
+      "epoch": 0.53515625,
+      "grad_norm": 0.16174718737602234,
+      "learning_rate": 8.894817284917364e-05,
+      "loss": 0.005412050988525152,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00543,
+      "step": 137,
+      "tokens/total": 4143344,
+      "tokens/train_per_sec_per_gpu": 35.39,
+      "tokens/trainable": 62649
+    },
+    {
+      "epoch": 0.5390625,
+      "grad_norm": 0.10317311435937881,
+      "learning_rate": 8.875691779131569e-05,
+      "loss": 0.002703815931454301,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00271,
+      "step": 138,
+      "tokens/total": 4173648,
+      "tokens/train_per_sec_per_gpu": 36.72,
+      "tokens/trainable": 63100
+    },
+    {
+      "epoch": 0.54296875,
+      "grad_norm": 0.23210155963897705,
+      "learning_rate": 8.856425797031829e-05,
+      "loss": 0.00892355665564537,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00896,
+      "step": 139,
+      "tokens/total": 4204080,
+      "tokens/train_per_sec_per_gpu": 37.69,
+      "tokens/trainable": 63588
+    },
+    {
+      "epoch": 0.546875,
+      "grad_norm": 0.4230027496814728,
+      "learning_rate": 8.837020140354295e-05,
+      "loss": 0.026184625923633575,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.42,
+      "memory/max_allocated (GiB)": 33.42,
+      "ppl": 1.02653,
+      "step": 140,
+      "tokens/total": 4232400,
+      "tokens/train_per_sec_per_gpu": 33.19,
+      "tokens/trainable": 64005
+    },
+    {
+      "epoch": 0.55078125,
+      "grad_norm": 0.5172422528266907,
+      "learning_rate": 8.817475616647554e-05,
+      "loss": 0.004686606116592884,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.0047,
+      "step": 141,
+      "tokens/total": 4262624,
+      "tokens/train_per_sec_per_gpu": 35.22,
+      "tokens/trainable": 64462
+    },
+    {
+      "epoch": 0.5546875,
+      "grad_norm": 0.28759023547172546,
+      "learning_rate": 8.797793039239017e-05,
+      "loss": 0.009430551901459694,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.00948,
+      "step": 142,
+      "tokens/total": 4293168,
+      "tokens/train_per_sec_per_gpu": 34.05,
+      "tokens/trainable": 64900
+    },
+    {
+      "epoch": 0.55859375,
+      "grad_norm": 0.45416516065597534,
+      "learning_rate": 8.777973227201069e-05,
+      "loss": 0.025679660961031914,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.02601,
+      "step": 143,
+      "tokens/total": 4323248,
+      "tokens/train_per_sec_per_gpu": 37.66,
+      "tokens/trainable": 65379
+    },
+    {
+      "epoch": 0.5625,
+      "grad_norm": 0.09341549873352051,
+      "learning_rate": 8.758017005316988e-05,
+      "loss": 0.0024609356187283993,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00246,
+      "step": 144,
+      "tokens/total": 4353344,
+      "tokens/train_per_sec_per_gpu": 36.79,
+      "tokens/trainable": 65845
+    },
+    {
+      "epoch": 0.56640625,
+      "grad_norm": 0.06811556965112686,
+      "learning_rate": 8.737925204046629e-05,
+      "loss": 0.00176865397952497,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00177,
+      "step": 145,
+      "tokens/total": 4383584,
+      "tokens/train_per_sec_per_gpu": 33.07,
+      "tokens/trainable": 66285
+    },
+    {
+      "epoch": 0.5703125,
+      "grad_norm": 0.3615569472312927,
+      "learning_rate": 8.717698659491851e-05,
+      "loss": 0.018001921474933624,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01816,
+      "step": 146,
+      "tokens/total": 4413856,
+      "tokens/train_per_sec_per_gpu": 31.66,
+      "tokens/trainable": 66736
+    },
+    {
+      "epoch": 0.57421875,
+      "grad_norm": 0.37582314014434814,
+      "learning_rate": 8.697338213361735e-05,
+      "loss": 0.01264512725174427,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01273,
+      "step": 147,
+      "tokens/total": 4444112,
+      "tokens/train_per_sec_per_gpu": 35.33,
+      "tokens/trainable": 67150
+    },
+    {
+      "epoch": 0.578125,
+      "grad_norm": 0.3228939473628998,
+      "learning_rate": 8.676844712937552e-05,
+      "loss": 0.009639455936849117,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00969,
+      "step": 148,
+      "tokens/total": 4474464,
+      "tokens/train_per_sec_per_gpu": 35.2,
+      "tokens/trainable": 67633
+    },
+    {
+      "epoch": 0.58203125,
+      "grad_norm": 0.44322797656059265,
+      "learning_rate": 8.656219011037509e-05,
+      "loss": 0.013246036134660244,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01333,
+      "step": 149,
+      "tokens/total": 4504880,
+      "tokens/train_per_sec_per_gpu": 33.95,
+      "tokens/trainable": 68077
+    },
+    {
+      "epoch": 0.5859375,
+      "grad_norm": 0.27842915058135986,
+      "learning_rate": 8.63546196598125e-05,
+      "loss": 0.007822944782674313,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00785,
+      "step": 150,
+      "tokens/total": 4534976,
+      "tokens/train_per_sec_per_gpu": 33.33,
+      "tokens/trainable": 68545
+    },
+    {
+      "epoch": 0.58984375,
+      "grad_norm": 0.20541682839393616,
+      "learning_rate": 8.614574441554145e-05,
+      "loss": 0.005034038331359625,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00505,
+      "step": 151,
+      "tokens/total": 4565088,
+      "tokens/train_per_sec_per_gpu": 30.33,
+      "tokens/trainable": 69011
+    },
+    {
+      "epoch": 0.59375,
+      "grad_norm": 0.33200570940971375,
+      "learning_rate": 8.593557306971349e-05,
+      "loss": 0.007599984761327505,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.00763,
+      "step": 152,
+      "tokens/total": 4595600,
+      "tokens/train_per_sec_per_gpu": 35.5,
+      "tokens/trainable": 69465
+    },
+    {
+      "epoch": 0.59765625,
+      "grad_norm": 0.47942468523979187,
+      "learning_rate": 8.572411436841618e-05,
+      "loss": 0.011957819573581219,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01203,
+      "step": 153,
+      "tokens/total": 4626032,
+      "tokens/train_per_sec_per_gpu": 34.03,
+      "tokens/trainable": 69928
+    },
+    {
+      "epoch": 0.6015625,
+      "grad_norm": 0.13327482342720032,
+      "learning_rate": 8.551137711130922e-05,
+      "loss": 0.0018434002995491028,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00185,
+      "step": 154,
+      "tokens/total": 4656352,
+      "tokens/train_per_sec_per_gpu": 36.8,
+      "tokens/trainable": 70395
+    },
+    {
+      "epoch": 0.60546875,
+      "grad_norm": 0.2294115424156189,
+      "learning_rate": 8.529737015125824e-05,
+      "loss": 0.003923698328435421,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00393,
+      "step": 155,
+      "tokens/total": 4684672,
+      "tokens/train_per_sec_per_gpu": 35.69,
+      "tokens/trainable": 70829
+    },
+    {
+      "epoch": 0.609375,
+      "grad_norm": 0.18192212283611298,
+      "learning_rate": 8.508210239396639e-05,
+      "loss": 0.0024503993336111307,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00245,
+      "step": 156,
+      "tokens/total": 4715216,
+      "tokens/train_per_sec_per_gpu": 34.05,
+      "tokens/trainable": 71287
+    },
+    {
+      "epoch": 0.61328125,
+      "grad_norm": 0.015309900976717472,
+      "learning_rate": 8.486558279760375e-05,
+      "loss": 0.0003224749234504998,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00032,
+      "step": 157,
+      "tokens/total": 4745376,
+      "tokens/train_per_sec_per_gpu": 35.1,
+      "tokens/trainable": 71749
+    },
+    {
+      "epoch": 0.6171875,
+      "grad_norm": 0.39558145403862,
+      "learning_rate": 8.464782037243449e-05,
+      "loss": 0.02732035145163536,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.0277,
+      "step": 158,
+      "tokens/total": 4775776,
+      "tokens/train_per_sec_per_gpu": 32.33,
+      "tokens/trainable": 72185
+    },
+    {
+      "epoch": 0.62109375,
+      "grad_norm": 0.2731363773345947,
+      "learning_rate": 8.442882418044202e-05,
+      "loss": 0.00566218001767993,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.00568,
+      "step": 159,
+      "tokens/total": 4805856,
+      "tokens/train_per_sec_per_gpu": 30.68,
+      "tokens/trainable": 72604
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 0.41349318623542786,
+      "learning_rate": 8.420860333495179e-05,
+      "loss": 0.00760249700397253,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00763,
+      "step": 160,
+      "tokens/total": 4836304,
+      "tokens/train_per_sec_per_gpu": 31.65,
+      "tokens/trainable": 73061
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 512,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 32,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.282682146024822e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-160/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b
+size 8273

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-192/README.md

@@ -0,0 +1,208 @@
+---
+base_model: /workspace/wave/parent
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- axolotl
+- base_model:adapter:/workspace/wave/parent
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-192/adapter_config.json

@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/workspace/wave/parent",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": null,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "down_proj",
+    "gate_proj",
+    "q_proj",
+    "o_proj",
+    "k_proj",
+    "up_proj"
+  ],
+  "target_parameters": [],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-192/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bb2d6e80c9b66d62b36ae40e54bbbcdf0c97db5c46489f1828b7a33386a0cd71
+size 547777976

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-192/chat_template.jinja

@@ -0,0 +1,47 @@
+{{ bos_token }}
+{%- if messages[0]['role'] == 'system' -%}
+    {%- if messages[0]['content'] is string -%}
+        {%- set first_user_prefix = messages[0]['content'] + '
+
+' -%}
+    {%- else -%}
+        {%- set first_user_prefix = messages[0]['content'][0]['text'] + '
+
+' -%}
+    {%- endif -%}
+    {%- set loop_messages = messages[1:] -%}
+{%- else -%}
+    {%- set first_user_prefix = "" -%}
+    {%- set loop_messages = messages -%}
+{%- endif -%}
+{%- for message in loop_messages -%}
+    {%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}
+        {{ raise_exception("Conversation roles must alternate user/assistant/user/assistant/...") }}
+    {%- endif -%}
+    {%- if (message['role'] == 'assistant') -%}
+        {%- set role = "model" -%}
+    {%- else -%}
+        {%- set role = message['role'] -%}
+    {%- endif -%}
+    {{ '<start_of_turn>' + role + '
+' + (first_user_prefix if loop.first else "") }}
+    {%- if message['content'] is string -%}
+        {{ message['content'] | trim }}
+    {%- elif message['content'] is iterable -%}
+        {%- for item in message['content'] -%}
+            {%- if item['type'] == 'image' -%}
+                {{ '<start_of_image>' }}
+            {%- elif item['type'] == 'text' -%}
+                {{ item['text'] | trim }}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- else -%}
+        {{ raise_exception("Invalid content type") }}
+    {%- endif -%}
+    {{ '<end_of_turn>
+' }}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{'<start_of_turn>model
+'}}
+{%- endif -%}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-192/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:669b3fb6f447d4c422509832d768e3d5b970e63da72ded60c0a929f4ae9cb1bf
+size 1048106435

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-192/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d69b2462378ed67f5f5555fb6b192965044d7652528913ba5ccc13c8e15172a
+size 14645

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-192/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:076826708eb68e7e0324ae251ebe1ef8facf8800fbc47028be05da96491371fd
+size 1465

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-192/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726
+size 33384567

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-192/tokenizer_config.json

@@ -0,0 +1,26 @@
+{
+  "backend": "tokenizers",
+  "boi_token": "<start_of_image>",
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eoi_token": "<end_of_image>",
+  "eos_token": "<eos>",
+  "image_token": "<image_soft_token>",
+  "is_local": false,
+  "local_files_only": false,
+  "mask_token": "<mask>",
+  "model_max_length": 131072,
+  "model_specific_special_tokens": {
+    "boi_token": "<start_of_image>",
+    "eoi_token": "<end_of_image>",
+    "image_token": "<image_soft_token>"
+  },
+  "pad_token": "<pad>",
+  "padding_side": "left",
+  "processor_class": "Gemma3Processor",
+  "sp_model_kwargs": null,
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-192/tokens_state.json

@@ -0,0 +1 @@
+{"total": 5807248, "trainable": 87736}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-192/trainer_state.json

@@ -0,0 +1,2722 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.75,
+  "eval_steps": 500,
+  "global_step": 192,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00390625,
+      "grad_norm": 0.9943057298660278,
+      "learning_rate": 0.0,
+      "loss": 0.11221420764923096,
+      "memory/device_reserved (GiB)": 33.9,
+      "memory/max_active (GiB)": 32.79,
+      "memory/max_allocated (GiB)": 32.79,
+      "ppl": 1.11875,
+      "step": 1,
+      "tokens/total": 30176,
+      "tokens/train_per_sec_per_gpu": 27.43,
+      "tokens/trainable": 417
+    },
+    {
+      "epoch": 0.0078125,
+      "grad_norm": 0.8786405920982361,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.11656901985406876,
+      "memory/device_reserved (GiB)": 34.68,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.12364,
+      "step": 2,
+      "tokens/total": 60272,
+      "tokens/train_per_sec_per_gpu": 34.84,
+      "tokens/trainable": 873
+    },
+    {
+      "epoch": 0.01171875,
+      "grad_norm": 0.8411452770233154,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.1128283143043518,
+      "memory/device_reserved (GiB)": 34.72,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.11944,
+      "step": 3,
+      "tokens/total": 90512,
+      "tokens/train_per_sec_per_gpu": 35.03,
+      "tokens/trainable": 1353
+    },
+    {
+      "epoch": 0.015625,
+      "grad_norm": 0.8970848917961121,
+      "learning_rate": 1.2e-05,
+      "loss": 0.10135132074356079,
+      "memory/device_reserved (GiB)": 34.73,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.10667,
+      "step": 4,
+      "tokens/total": 120944,
+      "tokens/train_per_sec_per_gpu": 31.32,
+      "tokens/trainable": 1778
+    },
+    {
+      "epoch": 0.01953125,
+      "grad_norm": 0.7805498242378235,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.11149105429649353,
+      "memory/device_reserved (GiB)": 35.12,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.11794,
+      "step": 5,
+      "tokens/total": 151440,
+      "tokens/train_per_sec_per_gpu": 36.88,
+      "tokens/trainable": 2261
+    },
+    {
+      "epoch": 0.0234375,
+      "grad_norm": 0.9325005412101746,
+      "learning_rate": 2e-05,
+      "loss": 0.09598484635353088,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.10074,
+      "step": 6,
+      "tokens/total": 181984,
+      "tokens/train_per_sec_per_gpu": 33.02,
+      "tokens/trainable": 2705
+    },
+    {
+      "epoch": 0.02734375,
+      "grad_norm": 0.9768337607383728,
+      "learning_rate": 2.4e-05,
+      "loss": 0.06007109954953194,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.06191,
+      "step": 7,
+      "tokens/total": 212336,
+      "tokens/train_per_sec_per_gpu": 34.05,
+      "tokens/trainable": 3136
+    },
+    {
+      "epoch": 0.03125,
+      "grad_norm": 1.0281821489334106,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.07311521470546722,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.07585,
+      "step": 8,
+      "tokens/total": 242592,
+      "tokens/train_per_sec_per_gpu": 34.29,
+      "tokens/trainable": 3579
+    },
+    {
+      "epoch": 0.03515625,
+      "grad_norm": 1.3234435319900513,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.044495582580566406,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.0455,
+      "step": 9,
+      "tokens/total": 272784,
+      "tokens/train_per_sec_per_gpu": 29.2,
+      "tokens/trainable": 4000
+    },
+    {
+      "epoch": 0.0390625,
+      "grad_norm": 1.8732832670211792,
+      "learning_rate": 3.6e-05,
+      "loss": 0.032459765672683716,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.03299,
+      "step": 10,
+      "tokens/total": 303184,
+      "tokens/train_per_sec_per_gpu": 37.86,
+      "tokens/trainable": 4474
+    },
+    {
+      "epoch": 0.04296875,
+      "grad_norm": 1.0495871305465698,
+      "learning_rate": 4e-05,
+      "loss": 0.02137758582830429,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.02161,
+      "step": 11,
+      "tokens/total": 333296,
+      "tokens/train_per_sec_per_gpu": 33.24,
+      "tokens/trainable": 4942
+    },
+    {
+      "epoch": 0.046875,
+      "grad_norm": 2.064997673034668,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.04079515486955643,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.04164,
+      "step": 12,
+      "tokens/total": 363840,
+      "tokens/train_per_sec_per_gpu": 34.35,
+      "tokens/trainable": 5424
+    },
+    {
+      "epoch": 0.05078125,
+      "grad_norm": 2.9026453495025635,
+      "learning_rate": 4.8e-05,
+      "loss": 0.06431213766336441,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.06643,
+      "step": 13,
+      "tokens/total": 394112,
+      "tokens/train_per_sec_per_gpu": 35.34,
+      "tokens/trainable": 5855
+    },
+    {
+      "epoch": 0.0546875,
+      "grad_norm": 2.4172568321228027,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.07758771628141403,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.08068,
+      "step": 14,
+      "tokens/total": 424656,
+      "tokens/train_per_sec_per_gpu": 34.9,
+      "tokens/trainable": 6344
+    },
+    {
+      "epoch": 0.05859375,
+      "grad_norm": 2.6533877849578857,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 0.0820830836892128,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.08555,
+      "step": 15,
+      "tokens/total": 455152,
+      "tokens/train_per_sec_per_gpu": 36.2,
+      "tokens/trainable": 6825
+    },
+    {
+      "epoch": 0.0625,
+      "grad_norm": 2.7390506267547607,
+      "learning_rate": 6e-05,
+      "loss": 0.053628768771886826,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.05509,
+      "step": 16,
+      "tokens/total": 485328,
+      "tokens/train_per_sec_per_gpu": 30.12,
+      "tokens/trainable": 7230
+    },
+    {
+      "epoch": 0.06640625,
+      "grad_norm": 0.8172504305839539,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.016706150025129318,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01685,
+      "step": 17,
+      "tokens/total": 515744,
+      "tokens/train_per_sec_per_gpu": 36.07,
+      "tokens/trainable": 7699
+    },
+    {
+      "epoch": 0.0703125,
+      "grad_norm": 0.7143669724464417,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.021398911252617836,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02163,
+      "step": 18,
+      "tokens/total": 546144,
+      "tokens/train_per_sec_per_gpu": 33.9,
+      "tokens/trainable": 8184
+    },
+    {
+      "epoch": 0.07421875,
+      "grad_norm": 2.224132537841797,
+      "learning_rate": 7.2e-05,
+      "loss": 0.016565855592489243,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.0167,
+      "step": 19,
+      "tokens/total": 576560,
+      "tokens/train_per_sec_per_gpu": 35.01,
+      "tokens/trainable": 8675
+    },
+    {
+      "epoch": 0.078125,
+      "grad_norm": 1.9165221452713013,
+      "learning_rate": 7.6e-05,
+      "loss": 0.038831885904073715,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.0396,
+      "step": 20,
+      "tokens/total": 607056,
+      "tokens/train_per_sec_per_gpu": 39.89,
+      "tokens/trainable": 9189
+    },
+    {
+      "epoch": 0.08203125,
+      "grad_norm": 0.9385339617729187,
+      "learning_rate": 8e-05,
+      "loss": 0.02326519787311554,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02354,
+      "step": 21,
+      "tokens/total": 637488,
+      "tokens/train_per_sec_per_gpu": 31.74,
+      "tokens/trainable": 9614
+    },
+    {
+      "epoch": 0.0859375,
+      "grad_norm": 1.3244878053665161,
+      "learning_rate": 8.4e-05,
+      "loss": 0.04611007124185562,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.04719,
+      "step": 22,
+      "tokens/total": 667632,
+      "tokens/train_per_sec_per_gpu": 36.23,
+      "tokens/trainable": 10086
+    },
+    {
+      "epoch": 0.08984375,
+      "grad_norm": 1.130301594734192,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.03128095716238022,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.03178,
+      "step": 23,
+      "tokens/total": 697872,
+      "tokens/train_per_sec_per_gpu": 32.92,
+      "tokens/trainable": 10535
+    },
+    {
+      "epoch": 0.09375,
+      "grad_norm": 1.9159126281738281,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 0.042277175933122635,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.04318,
+      "step": 24,
+      "tokens/total": 728432,
+      "tokens/train_per_sec_per_gpu": 31.63,
+      "tokens/trainable": 10993
+    },
+    {
+      "epoch": 0.09765625,
+      "grad_norm": 0.5571489334106445,
+      "learning_rate": 9.6e-05,
+      "loss": 0.025266434997320175,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02559,
+      "step": 25,
+      "tokens/total": 756656,
+      "tokens/train_per_sec_per_gpu": 40.08,
+      "tokens/trainable": 11417
+    },
+    {
+      "epoch": 0.1015625,
+      "grad_norm": 0.8746730089187622,
+      "learning_rate": 0.0001,
+      "loss": 0.022446369752287865,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.0227,
+      "step": 26,
+      "tokens/total": 787120,
+      "tokens/train_per_sec_per_gpu": 34.39,
+      "tokens/trainable": 11874
+    },
+    {
+      "epoch": 0.10546875,
+      "grad_norm": 1.505344271659851,
+      "learning_rate": 9.99990636831587e-05,
+      "loss": 0.02043886110186577,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02065,
+      "step": 27,
+      "tokens/total": 817504,
+      "tokens/train_per_sec_per_gpu": 31.74,
+      "tokens/trainable": 12316
+    },
+    {
+      "epoch": 0.109375,
+      "grad_norm": 1.0260790586471558,
+      "learning_rate": 9.999625477159879e-05,
+      "loss": 0.021617872640490532,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02185,
+      "step": 28,
+      "tokens/total": 848128,
+      "tokens/train_per_sec_per_gpu": 37.86,
+      "tokens/trainable": 12800
+    },
+    {
+      "epoch": 0.11328125,
+      "grad_norm": 0.8115366101264954,
+      "learning_rate": 9.999157338221051e-05,
+      "loss": 0.020092125982046127,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0203,
+      "step": 29,
+      "tokens/total": 878448,
+      "tokens/train_per_sec_per_gpu": 35.57,
+      "tokens/trainable": 13270
+    },
+    {
+      "epoch": 0.1171875,
+      "grad_norm": 0.5937158465385437,
+      "learning_rate": 9.998501970980562e-05,
+      "loss": 0.014808757230639458,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.01492,
+      "step": 30,
+      "tokens/total": 909008,
+      "tokens/train_per_sec_per_gpu": 35.43,
+      "tokens/trainable": 13752
+    },
+    {
+      "epoch": 0.12109375,
+      "grad_norm": 0.7298683524131775,
+      "learning_rate": 9.997659402710915e-05,
+      "loss": 0.019437074661254883,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01963,
+      "step": 31,
+      "tokens/total": 939312,
+      "tokens/train_per_sec_per_gpu": 34.12,
+      "tokens/trainable": 14218
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 0.17546068131923676,
+      "learning_rate": 9.996629668474818e-05,
+      "loss": 0.003190835705026984,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.0032,
+      "step": 32,
+      "tokens/total": 969264,
+      "tokens/train_per_sec_per_gpu": 34.52,
+      "tokens/trainable": 14655
+    },
+    {
+      "epoch": 0.12890625,
+      "grad_norm": 0.8119296431541443,
+      "learning_rate": 9.995412811123711e-05,
+      "loss": 0.021450746804475784,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.02168,
+      "step": 33,
+      "tokens/total": 999792,
+      "tokens/train_per_sec_per_gpu": 33.42,
+      "tokens/trainable": 15087
+    },
+    {
+      "epoch": 0.1328125,
+      "grad_norm": 0.7597190141677856,
+      "learning_rate": 9.994008881295999e-05,
+      "loss": 0.02241288125514984,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02267,
+      "step": 34,
+      "tokens/total": 1029840,
+      "tokens/train_per_sec_per_gpu": 34.0,
+      "tokens/trainable": 15548
+    },
+    {
+      "epoch": 0.13671875,
+      "grad_norm": 2.3171119689941406,
+      "learning_rate": 9.992417937414932e-05,
+      "loss": 0.021416299045085907,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.02165,
+      "step": 35,
+      "tokens/total": 1060416,
+      "tokens/train_per_sec_per_gpu": 33.79,
+      "tokens/trainable": 16037
+    },
+    {
+      "epoch": 0.140625,
+      "grad_norm": 0.4047625958919525,
+      "learning_rate": 9.99064004568618e-05,
+      "loss": 0.005595909897238016,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.72,
+      "memory/max_allocated (GiB)": 33.72,
+      "ppl": 1.00561,
+      "step": 36,
+      "tokens/total": 1090464,
+      "tokens/train_per_sec_per_gpu": 31.13,
+      "tokens/trainable": 16475
+    },
+    {
+      "epoch": 0.14453125,
+      "grad_norm": 0.5865066051483154,
+      "learning_rate": 9.988675280095074e-05,
+      "loss": 0.010679518803954124,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.01074,
+      "step": 37,
+      "tokens/total": 1120944,
+      "tokens/train_per_sec_per_gpu": 34.44,
+      "tokens/trainable": 16919
+    },
+    {
+      "epoch": 0.1484375,
+      "grad_norm": 1.583072304725647,
+      "learning_rate": 9.986523722403528e-05,
+      "loss": 0.02127654477953911,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0215,
+      "step": 38,
+      "tokens/total": 1151360,
+      "tokens/train_per_sec_per_gpu": 29.02,
+      "tokens/trainable": 17333
+    },
+    {
+      "epoch": 0.15234375,
+      "grad_norm": 1.30342698097229,
+      "learning_rate": 9.984185462146642e-05,
+      "loss": 0.025776617228984833,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.02611,
+      "step": 39,
+      "tokens/total": 1181728,
+      "tokens/train_per_sec_per_gpu": 36.2,
+      "tokens/trainable": 17806
+    },
+    {
+      "epoch": 0.15625,
+      "grad_norm": 1.1053258180618286,
+      "learning_rate": 9.98166059662897e-05,
+      "loss": 0.026606226339936256,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02696,
+      "step": 40,
+      "tokens/total": 1212064,
+      "tokens/train_per_sec_per_gpu": 35.67,
+      "tokens/trainable": 18290
+    },
+    {
+      "epoch": 0.16015625,
+      "grad_norm": 0.6078025102615356,
+      "learning_rate": 9.978949230920472e-05,
+      "loss": 0.017936185002326965,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.0181,
+      "step": 41,
+      "tokens/total": 1242592,
+      "tokens/train_per_sec_per_gpu": 32.05,
+      "tokens/trainable": 18747
+    },
+    {
+      "epoch": 0.1640625,
+      "grad_norm": 0.8376394510269165,
+      "learning_rate": 9.976051477852141e-05,
+      "loss": 0.02379559725522995,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02408,
+      "step": 42,
+      "tokens/total": 1272800,
+      "tokens/train_per_sec_per_gpu": 31.9,
+      "tokens/trainable": 19172
+    },
+    {
+      "epoch": 0.16796875,
+      "grad_norm": 0.4673662781715393,
+      "learning_rate": 9.972967458011312e-05,
+      "loss": 0.012735492549836636,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 34.04,
+      "memory/max_allocated (GiB)": 34.04,
+      "ppl": 1.01282,
+      "step": 43,
+      "tokens/total": 1303600,
+      "tokens/train_per_sec_per_gpu": 31.35,
+      "tokens/trainable": 19622
+    },
+    {
+      "epoch": 0.171875,
+      "grad_norm": 0.38821765780448914,
+      "learning_rate": 9.96969729973664e-05,
+      "loss": 0.008668653666973114,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00871,
+      "step": 44,
+      "tokens/total": 1333936,
+      "tokens/train_per_sec_per_gpu": 33.34,
+      "tokens/trainable": 20073
+    },
+    {
+      "epoch": 0.17578125,
+      "grad_norm": 0.7656640410423279,
+      "learning_rate": 9.966241139112754e-05,
+      "loss": 0.020857544615864754,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.02108,
+      "step": 45,
+      "tokens/total": 1364288,
+      "tokens/train_per_sec_per_gpu": 31.55,
+      "tokens/trainable": 20497
+    },
+    {
+      "epoch": 0.1796875,
+      "grad_norm": 0.6551951766014099,
+      "learning_rate": 9.96259911996461e-05,
+      "loss": 0.020156342536211014,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02036,
+      "step": 46,
+      "tokens/total": 1394368,
+      "tokens/train_per_sec_per_gpu": 34.35,
+      "tokens/trainable": 20940
+    },
+    {
+      "epoch": 0.18359375,
+      "grad_norm": 1.2237868309020996,
+      "learning_rate": 9.958771393851491e-05,
+      "loss": 0.015184836462140083,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.98,
+      "memory/max_allocated (GiB)": 33.98,
+      "ppl": 1.0153,
+      "step": 47,
+      "tokens/total": 1425024,
+      "tokens/train_per_sec_per_gpu": 35.07,
+      "tokens/trainable": 21405
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 1.571843147277832,
+      "learning_rate": 9.954758120060702e-05,
+      "loss": 0.01640632003545761,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01654,
+      "step": 48,
+      "tokens/total": 1455392,
+      "tokens/train_per_sec_per_gpu": 35.15,
+      "tokens/trainable": 21828
+    },
+    {
+      "epoch": 0.19140625,
+      "grad_norm": 1.3066037893295288,
+      "learning_rate": 9.950559465600948e-05,
+      "loss": 0.017745690420269966,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.0179,
+      "step": 49,
+      "tokens/total": 1485616,
+      "tokens/train_per_sec_per_gpu": 29.54,
+      "tokens/trainable": 22241
+    },
+    {
+      "epoch": 0.1953125,
+      "grad_norm": 0.6320858597755432,
+      "learning_rate": 9.946175605195379e-05,
+      "loss": 0.009900711476802826,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00995,
+      "step": 50,
+      "tokens/total": 1515984,
+      "tokens/train_per_sec_per_gpu": 36.38,
+      "tokens/trainable": 22688
+    },
+    {
+      "epoch": 0.19921875,
+      "grad_norm": 0.3268744647502899,
+      "learning_rate": 9.941606721274322e-05,
+      "loss": 0.006143633276224136,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00616,
+      "step": 51,
+      "tokens/total": 1546224,
+      "tokens/train_per_sec_per_gpu": 32.16,
+      "tokens/trainable": 23136
+    },
+    {
+      "epoch": 0.203125,
+      "grad_norm": 0.3406257629394531,
+      "learning_rate": 9.936853003967685e-05,
+      "loss": 0.004097095225006342,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00411,
+      "step": 52,
+      "tokens/total": 1576528,
+      "tokens/train_per_sec_per_gpu": 37.04,
+      "tokens/trainable": 23631
+    },
+    {
+      "epoch": 0.20703125,
+      "grad_norm": 1.1476560831069946,
+      "learning_rate": 9.93191465109705e-05,
+      "loss": 0.028221281245350838,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02862,
+      "step": 53,
+      "tokens/total": 1606768,
+      "tokens/train_per_sec_per_gpu": 34.12,
+      "tokens/trainable": 24096
+    },
+    {
+      "epoch": 0.2109375,
+      "grad_norm": 0.451351135969162,
+      "learning_rate": 9.926791868167438e-05,
+      "loss": 0.0078392019495368,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.66,
+      "memory/max_allocated (GiB)": 33.66,
+      "ppl": 1.00787,
+      "step": 54,
+      "tokens/total": 1636640,
+      "tokens/train_per_sec_per_gpu": 37.82,
+      "tokens/trainable": 24569
+    },
+    {
+      "epoch": 0.21484375,
+      "grad_norm": 0.8821313977241516,
+      "learning_rate": 9.921484868358753e-05,
+      "loss": 0.017667783424258232,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01782,
+      "step": 55,
+      "tokens/total": 1667136,
+      "tokens/train_per_sec_per_gpu": 34.02,
+      "tokens/trainable": 25067
+    },
+    {
+      "epoch": 0.21875,
+      "grad_norm": 0.6063156723976135,
+      "learning_rate": 9.915993872516924e-05,
+      "loss": 0.012062735855579376,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.01214,
+      "step": 56,
+      "tokens/total": 1697472,
+      "tokens/train_per_sec_per_gpu": 36.6,
+      "tokens/trainable": 25541
+    },
+    {
+      "epoch": 0.22265625,
+      "grad_norm": 0.7402717471122742,
+      "learning_rate": 9.9103191091447e-05,
+      "loss": 0.030087217688560486,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03054,
+      "step": 57,
+      "tokens/total": 1727680,
+      "tokens/train_per_sec_per_gpu": 33.28,
+      "tokens/trainable": 25965
+    },
+    {
+      "epoch": 0.2265625,
+      "grad_norm": 0.5932730436325073,
+      "learning_rate": 9.904460814392147e-05,
+      "loss": 0.013998140580952168,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0141,
+      "step": 58,
+      "tokens/total": 1758192,
+      "tokens/train_per_sec_per_gpu": 33.94,
+      "tokens/trainable": 26409
+    },
+    {
+      "epoch": 0.23046875,
+      "grad_norm": 0.6833471059799194,
+      "learning_rate": 9.898419232046825e-05,
+      "loss": 0.009387579746544361,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00943,
+      "step": 59,
+      "tokens/total": 1788816,
+      "tokens/train_per_sec_per_gpu": 34.07,
+      "tokens/trainable": 26883
+    },
+    {
+      "epoch": 0.234375,
+      "grad_norm": 0.49946627020835876,
+      "learning_rate": 9.892194613523633e-05,
+      "loss": 0.014185642823576927,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01429,
+      "step": 60,
+      "tokens/total": 1819056,
+      "tokens/train_per_sec_per_gpu": 33.72,
+      "tokens/trainable": 27323
+    },
+    {
+      "epoch": 0.23828125,
+      "grad_norm": 0.17088162899017334,
+      "learning_rate": 9.885787217854357e-05,
+      "loss": 0.003224549815058708,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00323,
+      "step": 61,
+      "tokens/total": 1849664,
+      "tokens/train_per_sec_per_gpu": 39.22,
+      "tokens/trainable": 27830
+    },
+    {
+      "epoch": 0.2421875,
+      "grad_norm": 0.5056981444358826,
+      "learning_rate": 9.879197311676887e-05,
+      "loss": 0.012983381748199463,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 34.02,
+      "memory/max_allocated (GiB)": 34.02,
+      "ppl": 1.01307,
+      "step": 62,
+      "tokens/total": 1880272,
+      "tokens/train_per_sec_per_gpu": 32.36,
+      "tokens/trainable": 28281
+    },
+    {
+      "epoch": 0.24609375,
+      "grad_norm": 2.0115866661071777,
+      "learning_rate": 9.872425169224113e-05,
+      "loss": 0.02842879481613636,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.02884,
+      "step": 63,
+      "tokens/total": 1910752,
+      "tokens/train_per_sec_per_gpu": 32.32,
+      "tokens/trainable": 28742
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8811206221580505,
+      "learning_rate": 9.865471072312528e-05,
+      "loss": 0.016607588157057762,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01675,
+      "step": 64,
+      "tokens/total": 1940848,
+      "tokens/train_per_sec_per_gpu": 33.58,
+      "tokens/trainable": 29183
+    },
+    {
+      "epoch": 0.25390625,
+      "grad_norm": 0.12346348166465759,
+      "learning_rate": 9.858335310330492e-05,
+      "loss": 0.003063689451664686,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00307,
+      "step": 65,
+      "tokens/total": 1971344,
+      "tokens/train_per_sec_per_gpu": 31.61,
+      "tokens/trainable": 29648
+    },
+    {
+      "epoch": 0.2578125,
+      "grad_norm": 0.28303098678588867,
+      "learning_rate": 9.851018180226185e-05,
+      "loss": 0.00448613939806819,
+      "memory/device_reserved (GiB)": 35.06,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.0045,
+      "step": 66,
+      "tokens/total": 2001712,
+      "tokens/train_per_sec_per_gpu": 33.54,
+      "tokens/trainable": 30075
+    },
+    {
+      "epoch": 0.26171875,
+      "grad_norm": 0.7989771962165833,
+      "learning_rate": 9.843519986495259e-05,
+      "loss": 0.008180832490324974,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.00821,
+      "step": 67,
+      "tokens/total": 2029936,
+      "tokens/train_per_sec_per_gpu": 37.66,
+      "tokens/trainable": 30546
+    },
+    {
+      "epoch": 0.265625,
+      "grad_norm": 0.12454555928707123,
+      "learning_rate": 9.835841041168162e-05,
+      "loss": 0.0016878236783668399,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00169,
+      "step": 68,
+      "tokens/total": 2060384,
+      "tokens/train_per_sec_per_gpu": 35.16,
+      "tokens/trainable": 31025
+    },
+    {
+      "epoch": 0.26953125,
+      "grad_norm": 1.7390296459197998,
+      "learning_rate": 9.82798166379715e-05,
+      "loss": 0.016503486782312393,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.01664,
+      "step": 69,
+      "tokens/total": 2090688,
+      "tokens/train_per_sec_per_gpu": 33.9,
+      "tokens/trainable": 31505
+    },
+    {
+      "epoch": 0.2734375,
+      "grad_norm": 0.968937337398529,
+      "learning_rate": 9.819942181443002e-05,
+      "loss": 0.01685403287410736,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.017,
+      "step": 70,
+      "tokens/total": 2120848,
+      "tokens/train_per_sec_per_gpu": 32.32,
+      "tokens/trainable": 31956
+    },
+    {
+      "epoch": 0.27734375,
+      "grad_norm": 0.3503779470920563,
+      "learning_rate": 9.811722928661392e-05,
+      "loss": 0.005242045037448406,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.35,
+      "memory/max_allocated (GiB)": 33.35,
+      "ppl": 1.00526,
+      "step": 71,
+      "tokens/total": 2149120,
+      "tokens/train_per_sec_per_gpu": 33.45,
+      "tokens/trainable": 32374
+    },
+    {
+      "epoch": 0.28125,
+      "grad_norm": 0.5608110427856445,
+      "learning_rate": 9.803324247488975e-05,
+      "loss": 0.013865578919649124,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01396,
+      "step": 72,
+      "tokens/total": 2179472,
+      "tokens/train_per_sec_per_gpu": 34.56,
+      "tokens/trainable": 32833
+    },
+    {
+      "epoch": 0.28515625,
+      "grad_norm": 1.2297694683074951,
+      "learning_rate": 9.794746487429161e-05,
+      "loss": 0.017295369878411293,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01745,
+      "step": 73,
+      "tokens/total": 2209984,
+      "tokens/train_per_sec_per_gpu": 35.72,
+      "tokens/trainable": 33304
+    },
+    {
+      "epoch": 0.2890625,
+      "grad_norm": 0.1584348827600479,
+      "learning_rate": 9.785990005437554e-05,
+      "loss": 0.004732954781502485,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00474,
+      "step": 74,
+      "tokens/total": 2240512,
+      "tokens/train_per_sec_per_gpu": 32.44,
+      "tokens/trainable": 33771
+    },
+    {
+      "epoch": 0.29296875,
+      "grad_norm": 0.3246576488018036,
+      "learning_rate": 9.777055165907117e-05,
+      "loss": 0.013291046023368835,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01338,
+      "step": 75,
+      "tokens/total": 2270608,
+      "tokens/train_per_sec_per_gpu": 34.64,
+      "tokens/trainable": 34186
+    },
+    {
+      "epoch": 0.296875,
+      "grad_norm": 1.1803282499313354,
+      "learning_rate": 9.767942340652993e-05,
+      "loss": 0.014760692603886127,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01487,
+      "step": 76,
+      "tokens/total": 2300864,
+      "tokens/train_per_sec_per_gpu": 34.96,
+      "tokens/trainable": 34667
+    },
+    {
+      "epoch": 0.30078125,
+      "grad_norm": 2.001532793045044,
+      "learning_rate": 9.758651908897035e-05,
+      "loss": 0.06761633604764938,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.76,
+      "memory/max_allocated (GiB)": 33.76,
+      "ppl": 1.06995,
+      "step": 77,
+      "tokens/total": 2330960,
+      "tokens/train_per_sec_per_gpu": 31.81,
+      "tokens/trainable": 35078
+    },
+    {
+      "epoch": 0.3046875,
+      "grad_norm": 0.29207468032836914,
+      "learning_rate": 9.749184257252033e-05,
+      "loss": 0.006066338159143925,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.00608,
+      "step": 78,
+      "tokens/total": 2361584,
+      "tokens/train_per_sec_per_gpu": 34.95,
+      "tokens/trainable": 35550
+    },
+    {
+      "epoch": 0.30859375,
+      "grad_norm": 0.5561455488204956,
+      "learning_rate": 9.739539779705614e-05,
+      "loss": 0.033555109053850174,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.03412,
+      "step": 79,
+      "tokens/total": 2391936,
+      "tokens/train_per_sec_per_gpu": 29.55,
+      "tokens/trainable": 35977
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 0.4088423550128937,
+      "learning_rate": 9.729718877603861e-05,
+      "loss": 0.014240232296288013,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01434,
+      "step": 80,
+      "tokens/total": 2422432,
+      "tokens/train_per_sec_per_gpu": 38.29,
+      "tokens/trainable": 36462
+    },
+    {
+      "epoch": 0.31640625,
+      "grad_norm": 0.2712189853191376,
+      "learning_rate": 9.719721959634592e-05,
+      "loss": 0.011144062504172325,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01121,
+      "step": 81,
+      "tokens/total": 2452752,
+      "tokens/train_per_sec_per_gpu": 33.24,
+      "tokens/trainable": 36919
+    },
+    {
+      "epoch": 0.3203125,
+      "grad_norm": 0.40229442715644836,
+      "learning_rate": 9.709549441810375e-05,
+      "loss": 0.012572492472827435,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.44,
+      "memory/max_allocated (GiB)": 33.44,
+      "ppl": 1.01265,
+      "step": 82,
+      "tokens/total": 2481216,
+      "tokens/train_per_sec_per_gpu": 33.48,
+      "tokens/trainable": 37379
+    },
+    {
+      "epoch": 0.32421875,
+      "grad_norm": 3.2695465087890625,
+      "learning_rate": 9.699201747451195e-05,
+      "loss": 0.012744108214974403,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01283,
+      "step": 83,
+      "tokens/total": 2511488,
+      "tokens/train_per_sec_per_gpu": 33.12,
+      "tokens/trainable": 37832
+    },
+    {
+      "epoch": 0.328125,
+      "grad_norm": 0.42963695526123047,
+      "learning_rate": 9.688679307166854e-05,
+      "loss": 0.014645008370280266,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01475,
+      "step": 84,
+      "tokens/total": 2541984,
+      "tokens/train_per_sec_per_gpu": 34.0,
+      "tokens/trainable": 38303
+    },
+    {
+      "epoch": 0.33203125,
+      "grad_norm": 0.6282569169998169,
+      "learning_rate": 9.677982558839042e-05,
+      "loss": 0.009028157219290733,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.69,
+      "memory/max_allocated (GiB)": 33.69,
+      "ppl": 1.00907,
+      "step": 85,
+      "tokens/total": 2572224,
+      "tokens/train_per_sec_per_gpu": 33.71,
+      "tokens/trainable": 38758
+    },
+    {
+      "epoch": 0.3359375,
+      "grad_norm": 0.5804964900016785,
+      "learning_rate": 9.66711194760312e-05,
+      "loss": 0.010930215008556843,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01099,
+      "step": 86,
+      "tokens/total": 2602704,
+      "tokens/train_per_sec_per_gpu": 30.71,
+      "tokens/trainable": 39179
+    },
+    {
+      "epoch": 0.33984375,
+      "grad_norm": 0.3565916121006012,
+      "learning_rate": 9.656067925829593e-05,
+      "loss": 0.011934653855860233,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01201,
+      "step": 87,
+      "tokens/total": 2633072,
+      "tokens/train_per_sec_per_gpu": 35.39,
+      "tokens/trainable": 39679
+    },
+    {
+      "epoch": 0.34375,
+      "grad_norm": 0.5799129009246826,
+      "learning_rate": 9.644850953105288e-05,
+      "loss": 0.010369446128606796,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.01042,
+      "step": 88,
+      "tokens/total": 2663232,
+      "tokens/train_per_sec_per_gpu": 32.76,
+      "tokens/trainable": 40118
+    },
+    {
+      "epoch": 0.34765625,
+      "grad_norm": 0.4171448051929474,
+      "learning_rate": 9.633461496214225e-05,
+      "loss": 0.012263461016118526,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.01234,
+      "step": 89,
+      "tokens/total": 2693696,
+      "tokens/train_per_sec_per_gpu": 32.95,
+      "tokens/trainable": 40571
+    },
+    {
+      "epoch": 0.3515625,
+      "grad_norm": 0.11327923834323883,
+      "learning_rate": 9.621900029118195e-05,
+      "loss": 0.0015271489974111319,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00153,
+      "step": 90,
+      "tokens/total": 2723984,
+      "tokens/train_per_sec_per_gpu": 30.97,
+      "tokens/trainable": 40997
+    },
+    {
+      "epoch": 0.35546875,
+      "grad_norm": 0.49398481845855713,
+      "learning_rate": 9.610167032937036e-05,
+      "loss": 0.0075433989986777306,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00757,
+      "step": 91,
+      "tokens/total": 2754240,
+      "tokens/train_per_sec_per_gpu": 36.54,
+      "tokens/trainable": 41462
+    },
+    {
+      "epoch": 0.359375,
+      "grad_norm": 0.3695297837257385,
+      "learning_rate": 9.598262995928611e-05,
+      "loss": 0.0067325979471206665,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00676,
+      "step": 92,
+      "tokens/total": 2784672,
+      "tokens/train_per_sec_per_gpu": 36.93,
+      "tokens/trainable": 41927
+    },
+    {
+      "epoch": 0.36328125,
+      "grad_norm": 0.6571913957595825,
+      "learning_rate": 9.586188413468492e-05,
+      "loss": 0.010379074141383171,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01043,
+      "step": 93,
+      "tokens/total": 2815120,
+      "tokens/train_per_sec_per_gpu": 34.92,
+      "tokens/trainable": 42397
+    },
+    {
+      "epoch": 0.3671875,
+      "grad_norm": 0.09913541376590729,
+      "learning_rate": 9.57394378802934e-05,
+      "loss": 0.0005878547672182322,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.00059,
+      "step": 94,
+      "tokens/total": 2845600,
+      "tokens/train_per_sec_per_gpu": 36.69,
+      "tokens/trainable": 42883
+    },
+    {
+      "epoch": 0.37109375,
+      "grad_norm": 0.3970830738544464,
+      "learning_rate": 9.56152962916e-05,
+      "loss": 0.005916361231356859,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00593,
+      "step": 95,
+      "tokens/total": 2875952,
+      "tokens/train_per_sec_per_gpu": 38.83,
+      "tokens/trainable": 43381
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 0.1583622545003891,
+      "learning_rate": 9.548946453464296e-05,
+      "loss": 0.0020841225050389767,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00209,
+      "step": 96,
+      "tokens/total": 2906288,
+      "tokens/train_per_sec_per_gpu": 33.1,
+      "tokens/trainable": 43824
+    },
+    {
+      "epoch": 0.37890625,
+      "grad_norm": 0.42756563425064087,
+      "learning_rate": 9.53619478457953e-05,
+      "loss": 0.004547779448330402,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00456,
+      "step": 97,
+      "tokens/total": 2936752,
+      "tokens/train_per_sec_per_gpu": 31.67,
+      "tokens/trainable": 44255
+    },
+    {
+      "epoch": 0.3828125,
+      "grad_norm": 0.9736410975456238,
+      "learning_rate": 9.523275153154695e-05,
+      "loss": 0.04175024852156639,
+      "memory/device_reserved (GiB)": 35.19,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.04263,
+      "step": 98,
+      "tokens/total": 2967072,
+      "tokens/train_per_sec_per_gpu": 37.88,
+      "tokens/trainable": 44739
+    },
+    {
+      "epoch": 0.38671875,
+      "grad_norm": 0.8381142020225525,
+      "learning_rate": 9.51018809682839e-05,
+      "loss": 0.013025594875216484,
+      "memory/device_reserved (GiB)": 35.76,
+      "memory/max_active (GiB)": 34.03,
+      "memory/max_allocated (GiB)": 34.03,
+      "ppl": 1.01311,
+      "step": 99,
+      "tokens/total": 2997664,
+      "tokens/train_per_sec_per_gpu": 36.31,
+      "tokens/trainable": 45202
+    },
+    {
+      "epoch": 0.390625,
+      "grad_norm": 0.22207525372505188,
+      "learning_rate": 9.49693416020645e-05,
+      "loss": 0.004689273424446583,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.0047,
+      "step": 100,
+      "tokens/total": 3028032,
+      "tokens/train_per_sec_per_gpu": 33.14,
+      "tokens/trainable": 45646
+    },
+    {
+      "epoch": 0.39453125,
+      "grad_norm": 0.28477466106414795,
+      "learning_rate": 9.483513894839276e-05,
+      "loss": 0.010324095375835896,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01038,
+      "step": 101,
+      "tokens/total": 3058272,
+      "tokens/train_per_sec_per_gpu": 33.09,
+      "tokens/trainable": 46089
+    },
+    {
+      "epoch": 0.3984375,
+      "grad_norm": 0.5098783373832703,
+      "learning_rate": 9.469927859198888e-05,
+      "loss": 0.028538472950458527,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.02895,
+      "step": 102,
+      "tokens/total": 3088848,
+      "tokens/train_per_sec_per_gpu": 32.63,
+      "tokens/trainable": 46547
+    },
+    {
+      "epoch": 0.40234375,
+      "grad_norm": 0.3906193673610687,
+      "learning_rate": 9.456176618655689e-05,
+      "loss": 0.017888592556118965,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.01805,
+      "step": 103,
+      "tokens/total": 3119424,
+      "tokens/train_per_sec_per_gpu": 32.02,
+      "tokens/trainable": 47014
+    },
+    {
+      "epoch": 0.40625,
+      "grad_norm": 0.5902650952339172,
+      "learning_rate": 9.442260745454927e-05,
+      "loss": 0.03132036700844765,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.03182,
+      "step": 104,
+      "tokens/total": 3149696,
+      "tokens/train_per_sec_per_gpu": 37.8,
+      "tokens/trainable": 47503
+    },
+    {
+      "epoch": 0.41015625,
+      "grad_norm": 0.2547222375869751,
+      "learning_rate": 9.428180818692884e-05,
+      "loss": 0.014553382061421871,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01466,
+      "step": 105,
+      "tokens/total": 3180064,
+      "tokens/train_per_sec_per_gpu": 36.37,
+      "tokens/trainable": 47996
+    },
+    {
+      "epoch": 0.4140625,
+      "grad_norm": 0.49491167068481445,
+      "learning_rate": 9.413937424292791e-05,
+      "loss": 0.0295359306037426,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.02998,
+      "step": 106,
+      "tokens/total": 3208320,
+      "tokens/train_per_sec_per_gpu": 34.18,
+      "tokens/trainable": 48447
+    },
+    {
+      "epoch": 0.41796875,
+      "grad_norm": 0.481608510017395,
+      "learning_rate": 9.399531154980424e-05,
+      "loss": 0.017734361812472343,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01789,
+      "step": 107,
+      "tokens/total": 3238416,
+      "tokens/train_per_sec_per_gpu": 30.78,
+      "tokens/trainable": 48889
+    },
+    {
+      "epoch": 0.421875,
+      "grad_norm": 0.3913033604621887,
+      "learning_rate": 9.384962610259455e-05,
+      "loss": 0.022145511582493782,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02239,
+      "step": 108,
+      "tokens/total": 3268784,
+      "tokens/train_per_sec_per_gpu": 38.03,
+      "tokens/trainable": 49360
+    },
+    {
+      "epoch": 0.42578125,
+      "grad_norm": 0.1751519739627838,
+      "learning_rate": 9.370232396386494e-05,
+      "loss": 0.0074960459023714066,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00752,
+      "step": 109,
+      "tokens/total": 3298912,
+      "tokens/train_per_sec_per_gpu": 36.89,
+      "tokens/trainable": 49827
+    },
+    {
+      "epoch": 0.4296875,
+      "grad_norm": 0.28872016072273254,
+      "learning_rate": 9.355341126345868e-05,
+      "loss": 0.01847866177558899,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.01865,
+      "step": 110,
+      "tokens/total": 3329232,
+      "tokens/train_per_sec_per_gpu": 34.36,
+      "tokens/trainable": 50267
+    },
+    {
+      "epoch": 0.43359375,
+      "grad_norm": 0.3326593041419983,
+      "learning_rate": 9.340289419824107e-05,
+      "loss": 0.013967925682663918,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.01407,
+      "step": 111,
+      "tokens/total": 3359440,
+      "tokens/train_per_sec_per_gpu": 33.47,
+      "tokens/trainable": 50711
+    },
+    {
+      "epoch": 0.4375,
+      "grad_norm": 0.47660934925079346,
+      "learning_rate": 9.325077903184159e-05,
+      "loss": 0.02260858193039894,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.02287,
+      "step": 112,
+      "tokens/total": 3389472,
+      "tokens/train_per_sec_per_gpu": 31.72,
+      "tokens/trainable": 51157
+    },
+    {
+      "epoch": 0.44140625,
+      "grad_norm": 0.3481523096561432,
+      "learning_rate": 9.30970720943932e-05,
+      "loss": 0.008511725813150406,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00855,
+      "step": 113,
+      "tokens/total": 3419920,
+      "tokens/train_per_sec_per_gpu": 32.79,
+      "tokens/trainable": 51613
+    },
+    {
+      "epoch": 0.4453125,
+      "grad_norm": 0.4288201928138733,
+      "learning_rate": 9.2941779782269e-05,
+      "loss": 0.02450375072658062,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.02481,
+      "step": 114,
+      "tokens/total": 3450176,
+      "tokens/train_per_sec_per_gpu": 29.19,
+      "tokens/trainable": 52017
+    },
+    {
+      "epoch": 0.44921875,
+      "grad_norm": 0.17284883558750153,
+      "learning_rate": 9.278490855781596e-05,
+      "loss": 0.007882107980549335,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00791,
+      "step": 115,
+      "tokens/total": 3480544,
+      "tokens/train_per_sec_per_gpu": 36.36,
+      "tokens/trainable": 52506
+    },
+    {
+      "epoch": 0.453125,
+      "grad_norm": 0.2693459689617157,
+      "learning_rate": 9.262646494908604e-05,
+      "loss": 0.005466018337756395,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00548,
+      "step": 116,
+      "tokens/total": 3510848,
+      "tokens/train_per_sec_per_gpu": 38.77,
+      "tokens/trainable": 53008
+    },
+    {
+      "epoch": 0.45703125,
+      "grad_norm": 0.48690056800842285,
+      "learning_rate": 9.246645554956457e-05,
+      "loss": 0.004998547025024891,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.00501,
+      "step": 117,
+      "tokens/total": 3541296,
+      "tokens/train_per_sec_per_gpu": 35.47,
+      "tokens/trainable": 53467
+    },
+    {
+      "epoch": 0.4609375,
+      "grad_norm": 0.43558549880981445,
+      "learning_rate": 9.230488701789578e-05,
+      "loss": 0.017219291999936104,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01737,
+      "step": 118,
+      "tokens/total": 3571936,
+      "tokens/train_per_sec_per_gpu": 32.96,
+      "tokens/trainable": 53923
+    },
+    {
+      "epoch": 0.46484375,
+      "grad_norm": 0.28566548228263855,
+      "learning_rate": 9.214176607760577e-05,
+      "loss": 0.00544874370098114,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00546,
+      "step": 119,
+      "tokens/total": 3602000,
+      "tokens/train_per_sec_per_gpu": 35.72,
+      "tokens/trainable": 54363
+    },
+    {
+      "epoch": 0.46875,
+      "grad_norm": 0.07611701637506485,
+      "learning_rate": 9.197709951682268e-05,
+      "loss": 0.0014362478395923972,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00144,
+      "step": 120,
+      "tokens/total": 3632288,
+      "tokens/train_per_sec_per_gpu": 35.29,
+      "tokens/trainable": 54839
+    },
+    {
+      "epoch": 0.47265625,
+      "grad_norm": 0.13201506435871124,
+      "learning_rate": 9.181089418799428e-05,
+      "loss": 0.0024022101424634457,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00241,
+      "step": 121,
+      "tokens/total": 3662608,
+      "tokens/train_per_sec_per_gpu": 32.7,
+      "tokens/trainable": 55267
+    },
+    {
+      "epoch": 0.4765625,
+      "grad_norm": 0.40975725650787354,
+      "learning_rate": 9.164315700760271e-05,
+      "loss": 0.008171994239091873,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.00821,
+      "step": 122,
+      "tokens/total": 3692864,
+      "tokens/train_per_sec_per_gpu": 30.6,
+      "tokens/trainable": 55700
+    },
+    {
+      "epoch": 0.48046875,
+      "grad_norm": 0.15150536596775055,
+      "learning_rate": 9.147389495587671e-05,
+      "loss": 0.0036906469613313675,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.0037,
+      "step": 123,
+      "tokens/total": 3722864,
+      "tokens/train_per_sec_per_gpu": 33.48,
+      "tokens/trainable": 56144
+    },
+    {
+      "epoch": 0.484375,
+      "grad_norm": 0.514573335647583,
+      "learning_rate": 9.130311507650116e-05,
+      "loss": 0.01570757105946541,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01583,
+      "step": 124,
+      "tokens/total": 3753056,
+      "tokens/train_per_sec_per_gpu": 33.18,
+      "tokens/trainable": 56589
+    },
+    {
+      "epoch": 0.48828125,
+      "grad_norm": 0.43223315477371216,
+      "learning_rate": 9.113082447632394e-05,
+      "loss": 0.007733152247965336,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00776,
+      "step": 125,
+      "tokens/total": 3783312,
+      "tokens/train_per_sec_per_gpu": 36.78,
+      "tokens/trainable": 57081
+    },
+    {
+      "epoch": 0.4921875,
+      "grad_norm": 0.2972177565097809,
+      "learning_rate": 9.09570303250602e-05,
+      "loss": 0.008276824839413166,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00831,
+      "step": 126,
+      "tokens/total": 3813712,
+      "tokens/train_per_sec_per_gpu": 32.97,
+      "tokens/trainable": 57527
+    },
+    {
+      "epoch": 0.49609375,
+      "grad_norm": 0.6973468065261841,
+      "learning_rate": 9.078173985499394e-05,
+      "loss": 0.011385848745703697,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.01145,
+      "step": 127,
+      "tokens/total": 3844336,
+      "tokens/train_per_sec_per_gpu": 34.98,
+      "tokens/trainable": 58005
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.11440204083919525,
+      "learning_rate": 9.060496036067713e-05,
+      "loss": 0.0026649129576981068,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.00267,
+      "step": 128,
+      "tokens/total": 3872688,
+      "tokens/train_per_sec_per_gpu": 35.34,
+      "tokens/trainable": 58454
+    },
+    {
+      "epoch": 0.50390625,
+      "grad_norm": 0.4737952649593353,
+      "learning_rate": 9.042669919862615e-05,
+      "loss": 0.011923262849450111,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.64,
+      "memory/max_allocated (GiB)": 33.64,
+      "ppl": 1.01199,
+      "step": 129,
+      "tokens/total": 3902640,
+      "tokens/train_per_sec_per_gpu": 32.11,
+      "tokens/trainable": 58896
+    },
+    {
+      "epoch": 0.5078125,
+      "grad_norm": 0.22904613614082336,
+      "learning_rate": 9.024696378701557e-05,
+      "loss": 0.003364444011822343,
+      "memory/device_reserved (GiB)": 34.68,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00337,
+      "step": 130,
+      "tokens/total": 3933104,
+      "tokens/train_per_sec_per_gpu": 35.96,
+      "tokens/trainable": 59356
+    },
+    {
+      "epoch": 0.51171875,
+      "grad_norm": 0.6996368169784546,
+      "learning_rate": 9.006576160536948e-05,
+      "loss": 0.0145557951182127,
+      "memory/device_reserved (GiB)": 34.92,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.01466,
+      "step": 131,
+      "tokens/total": 3963184,
+      "tokens/train_per_sec_per_gpu": 32.33,
+      "tokens/trainable": 59792
+    },
+    {
+      "epoch": 0.515625,
+      "grad_norm": 0.18643876910209656,
+      "learning_rate": 8.988310019425035e-05,
+      "loss": 0.004168116021901369,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00418,
+      "step": 132,
+      "tokens/total": 3993648,
+      "tokens/train_per_sec_per_gpu": 35.79,
+      "tokens/trainable": 60266
+    },
+    {
+      "epoch": 0.51953125,
+      "grad_norm": 0.3754265606403351,
+      "learning_rate": 8.969898715494506e-05,
+      "loss": 0.01239765528589487,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01247,
+      "step": 133,
+      "tokens/total": 4024000,
+      "tokens/train_per_sec_per_gpu": 35.46,
+      "tokens/trainable": 60759
+    },
+    {
+      "epoch": 0.5234375,
+      "grad_norm": 0.5720171332359314,
+      "learning_rate": 8.951343014914869e-05,
+      "loss": 0.028643302619457245,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.02906,
+      "step": 134,
+      "tokens/total": 4052480,
+      "tokens/train_per_sec_per_gpu": 31.88,
+      "tokens/trainable": 61204
+    },
+    {
+      "epoch": 0.52734375,
+      "grad_norm": 0.21384330093860626,
+      "learning_rate": 8.932643689864568e-05,
+      "loss": 0.005144410766661167,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00516,
+      "step": 135,
+      "tokens/total": 4082864,
+      "tokens/train_per_sec_per_gpu": 36.1,
+      "tokens/trainable": 61684
+    },
+    {
+      "epoch": 0.53125,
+      "grad_norm": 0.3025348484516144,
+      "learning_rate": 8.913801518498845e-05,
+      "loss": 0.011437858454883099,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.7,
+      "memory/max_allocated (GiB)": 33.7,
+      "ppl": 1.0115,
+      "step": 136,
+      "tokens/total": 4113040,
+      "tokens/train_per_sec_per_gpu": 38.36,
+      "tokens/trainable": 62159
+    },
+    {
+      "epoch": 0.53515625,
+      "grad_norm": 0.16174718737602234,
+      "learning_rate": 8.894817284917364e-05,
+      "loss": 0.005412050988525152,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00543,
+      "step": 137,
+      "tokens/total": 4143344,
+      "tokens/train_per_sec_per_gpu": 35.39,
+      "tokens/trainable": 62649
+    },
+    {
+      "epoch": 0.5390625,
+      "grad_norm": 0.10317311435937881,
+      "learning_rate": 8.875691779131569e-05,
+      "loss": 0.002703815931454301,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00271,
+      "step": 138,
+      "tokens/total": 4173648,
+      "tokens/train_per_sec_per_gpu": 36.72,
+      "tokens/trainable": 63100
+    },
+    {
+      "epoch": 0.54296875,
+      "grad_norm": 0.23210155963897705,
+      "learning_rate": 8.856425797031829e-05,
+      "loss": 0.00892355665564537,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00896,
+      "step": 139,
+      "tokens/total": 4204080,
+      "tokens/train_per_sec_per_gpu": 37.69,
+      "tokens/trainable": 63588
+    },
+    {
+      "epoch": 0.546875,
+      "grad_norm": 0.4230027496814728,
+      "learning_rate": 8.837020140354295e-05,
+      "loss": 0.026184625923633575,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.42,
+      "memory/max_allocated (GiB)": 33.42,
+      "ppl": 1.02653,
+      "step": 140,
+      "tokens/total": 4232400,
+      "tokens/train_per_sec_per_gpu": 33.19,
+      "tokens/trainable": 64005
+    },
+    {
+      "epoch": 0.55078125,
+      "grad_norm": 0.5172422528266907,
+      "learning_rate": 8.817475616647554e-05,
+      "loss": 0.004686606116592884,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.0047,
+      "step": 141,
+      "tokens/total": 4262624,
+      "tokens/train_per_sec_per_gpu": 35.22,
+      "tokens/trainable": 64462
+    },
+    {
+      "epoch": 0.5546875,
+      "grad_norm": 0.28759023547172546,
+      "learning_rate": 8.797793039239017e-05,
+      "loss": 0.009430551901459694,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.00948,
+      "step": 142,
+      "tokens/total": 4293168,
+      "tokens/train_per_sec_per_gpu": 34.05,
+      "tokens/trainable": 64900
+    },
+    {
+      "epoch": 0.55859375,
+      "grad_norm": 0.45416516065597534,
+      "learning_rate": 8.777973227201069e-05,
+      "loss": 0.025679660961031914,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.02601,
+      "step": 143,
+      "tokens/total": 4323248,
+      "tokens/train_per_sec_per_gpu": 37.66,
+      "tokens/trainable": 65379
+    },
+    {
+      "epoch": 0.5625,
+      "grad_norm": 0.09341549873352051,
+      "learning_rate": 8.758017005316988e-05,
+      "loss": 0.0024609356187283993,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00246,
+      "step": 144,
+      "tokens/total": 4353344,
+      "tokens/train_per_sec_per_gpu": 36.79,
+      "tokens/trainable": 65845
+    },
+    {
+      "epoch": 0.56640625,
+      "grad_norm": 0.06811556965112686,
+      "learning_rate": 8.737925204046629e-05,
+      "loss": 0.00176865397952497,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00177,
+      "step": 145,
+      "tokens/total": 4383584,
+      "tokens/train_per_sec_per_gpu": 33.07,
+      "tokens/trainable": 66285
+    },
+    {
+      "epoch": 0.5703125,
+      "grad_norm": 0.3615569472312927,
+      "learning_rate": 8.717698659491851e-05,
+      "loss": 0.018001921474933624,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01816,
+      "step": 146,
+      "tokens/total": 4413856,
+      "tokens/train_per_sec_per_gpu": 31.66,
+      "tokens/trainable": 66736
+    },
+    {
+      "epoch": 0.57421875,
+      "grad_norm": 0.37582314014434814,
+      "learning_rate": 8.697338213361735e-05,
+      "loss": 0.01264512725174427,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01273,
+      "step": 147,
+      "tokens/total": 4444112,
+      "tokens/train_per_sec_per_gpu": 35.33,
+      "tokens/trainable": 67150
+    },
+    {
+      "epoch": 0.578125,
+      "grad_norm": 0.3228939473628998,
+      "learning_rate": 8.676844712937552e-05,
+      "loss": 0.009639455936849117,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00969,
+      "step": 148,
+      "tokens/total": 4474464,
+      "tokens/train_per_sec_per_gpu": 35.2,
+      "tokens/trainable": 67633
+    },
+    {
+      "epoch": 0.58203125,
+      "grad_norm": 0.44322797656059265,
+      "learning_rate": 8.656219011037509e-05,
+      "loss": 0.013246036134660244,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01333,
+      "step": 149,
+      "tokens/total": 4504880,
+      "tokens/train_per_sec_per_gpu": 33.95,
+      "tokens/trainable": 68077
+    },
+    {
+      "epoch": 0.5859375,
+      "grad_norm": 0.27842915058135986,
+      "learning_rate": 8.63546196598125e-05,
+      "loss": 0.007822944782674313,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00785,
+      "step": 150,
+      "tokens/total": 4534976,
+      "tokens/train_per_sec_per_gpu": 33.33,
+      "tokens/trainable": 68545
+    },
+    {
+      "epoch": 0.58984375,
+      "grad_norm": 0.20541682839393616,
+      "learning_rate": 8.614574441554145e-05,
+      "loss": 0.005034038331359625,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00505,
+      "step": 151,
+      "tokens/total": 4565088,
+      "tokens/train_per_sec_per_gpu": 30.33,
+      "tokens/trainable": 69011
+    },
+    {
+      "epoch": 0.59375,
+      "grad_norm": 0.33200570940971375,
+      "learning_rate": 8.593557306971349e-05,
+      "loss": 0.007599984761327505,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.00763,
+      "step": 152,
+      "tokens/total": 4595600,
+      "tokens/train_per_sec_per_gpu": 35.5,
+      "tokens/trainable": 69465
+    },
+    {
+      "epoch": 0.59765625,
+      "grad_norm": 0.47942468523979187,
+      "learning_rate": 8.572411436841618e-05,
+      "loss": 0.011957819573581219,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01203,
+      "step": 153,
+      "tokens/total": 4626032,
+      "tokens/train_per_sec_per_gpu": 34.03,
+      "tokens/trainable": 69928
+    },
+    {
+      "epoch": 0.6015625,
+      "grad_norm": 0.13327482342720032,
+      "learning_rate": 8.551137711130922e-05,
+      "loss": 0.0018434002995491028,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00185,
+      "step": 154,
+      "tokens/total": 4656352,
+      "tokens/train_per_sec_per_gpu": 36.8,
+      "tokens/trainable": 70395
+    },
+    {
+      "epoch": 0.60546875,
+      "grad_norm": 0.2294115424156189,
+      "learning_rate": 8.529737015125824e-05,
+      "loss": 0.003923698328435421,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00393,
+      "step": 155,
+      "tokens/total": 4684672,
+      "tokens/train_per_sec_per_gpu": 35.69,
+      "tokens/trainable": 70829
+    },
+    {
+      "epoch": 0.609375,
+      "grad_norm": 0.18192212283611298,
+      "learning_rate": 8.508210239396639e-05,
+      "loss": 0.0024503993336111307,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00245,
+      "step": 156,
+      "tokens/total": 4715216,
+      "tokens/train_per_sec_per_gpu": 34.05,
+      "tokens/trainable": 71287
+    },
+    {
+      "epoch": 0.61328125,
+      "grad_norm": 0.015309900976717472,
+      "learning_rate": 8.486558279760375e-05,
+      "loss": 0.0003224749234504998,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00032,
+      "step": 157,
+      "tokens/total": 4745376,
+      "tokens/train_per_sec_per_gpu": 35.1,
+      "tokens/trainable": 71749
+    },
+    {
+      "epoch": 0.6171875,
+      "grad_norm": 0.39558145403862,
+      "learning_rate": 8.464782037243449e-05,
+      "loss": 0.02732035145163536,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.0277,
+      "step": 158,
+      "tokens/total": 4775776,
+      "tokens/train_per_sec_per_gpu": 32.33,
+      "tokens/trainable": 72185
+    },
+    {
+      "epoch": 0.62109375,
+      "grad_norm": 0.2731363773345947,
+      "learning_rate": 8.442882418044202e-05,
+      "loss": 0.00566218001767993,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.00568,
+      "step": 159,
+      "tokens/total": 4805856,
+      "tokens/train_per_sec_per_gpu": 30.68,
+      "tokens/trainable": 72604
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 0.41349318623542786,
+      "learning_rate": 8.420860333495179e-05,
+      "loss": 0.00760249700397253,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00763,
+      "step": 160,
+      "tokens/total": 4836304,
+      "tokens/train_per_sec_per_gpu": 31.65,
+      "tokens/trainable": 73061
+    },
+    {
+      "epoch": 0.62890625,
+      "grad_norm": 0.1329299956560135,
+      "learning_rate": 8.398716700025208e-05,
+      "loss": 0.002723761135712266,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.00273,
+      "step": 161,
+      "tokens/total": 4866544,
+      "tokens/train_per_sec_per_gpu": 30.1,
+      "tokens/trainable": 73474
+    },
+    {
+      "epoch": 0.6328125,
+      "grad_norm": 0.2733931839466095,
+      "learning_rate": 8.376452439121266e-05,
+      "loss": 0.008426842279732227,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.00846,
+      "step": 162,
+      "tokens/total": 4897120,
+      "tokens/train_per_sec_per_gpu": 32.76,
+      "tokens/trainable": 73923
+    },
+    {
+      "epoch": 0.63671875,
+      "grad_norm": 0.6801484227180481,
+      "learning_rate": 8.354068477290124e-05,
+      "loss": 0.02508145570755005,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.0254,
+      "step": 163,
+      "tokens/total": 4927440,
+      "tokens/train_per_sec_per_gpu": 35.46,
+      "tokens/trainable": 74379
+    },
+    {
+      "epoch": 0.640625,
+      "grad_norm": 0.22900699079036713,
+      "learning_rate": 8.331565746019807e-05,
+      "loss": 0.006623784080147743,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00665,
+      "step": 164,
+      "tokens/total": 4958032,
+      "tokens/train_per_sec_per_gpu": 33.52,
+      "tokens/trainable": 74830
+    },
+    {
+      "epoch": 0.64453125,
+      "grad_norm": 0.1291102021932602,
+      "learning_rate": 8.308945181740812e-05,
+      "loss": 0.0024572773836553097,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00246,
+      "step": 165,
+      "tokens/total": 4988368,
+      "tokens/train_per_sec_per_gpu": 31.84,
+      "tokens/trainable": 75262
+    },
+    {
+      "epoch": 0.6484375,
+      "grad_norm": 0.12582950294017792,
+      "learning_rate": 8.286207725787153e-05,
+      "loss": 0.0014331504935398698,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.00143,
+      "step": 166,
+      "tokens/total": 5018672,
+      "tokens/train_per_sec_per_gpu": 31.93,
+      "tokens/trainable": 75703
+    },
+    {
+      "epoch": 0.65234375,
+      "grad_norm": 0.10334278643131256,
+      "learning_rate": 8.263354324357182e-05,
+      "loss": 0.003322131698951125,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.00333,
+      "step": 167,
+      "tokens/total": 5049344,
+      "tokens/train_per_sec_per_gpu": 31.36,
+      "tokens/trainable": 76164
+    },
+    {
+      "epoch": 0.65625,
+      "grad_norm": 0.344831258058548,
+      "learning_rate": 8.240385928474219e-05,
+      "loss": 0.01979166269302368,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.01999,
+      "step": 168,
+      "tokens/total": 5079856,
+      "tokens/train_per_sec_per_gpu": 34.58,
+      "tokens/trainable": 76623
+    },
+    {
+      "epoch": 0.66015625,
+      "grad_norm": 0.3123904764652252,
+      "learning_rate": 8.217303493946967e-05,
+      "loss": 0.020188894122838974,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02039,
+      "step": 169,
+      "tokens/total": 5110192,
+      "tokens/train_per_sec_per_gpu": 34.46,
+      "tokens/trainable": 77047
+    },
+    {
+      "epoch": 0.6640625,
+      "grad_norm": 0.24240995943546295,
+      "learning_rate": 8.194107981329746e-05,
+      "loss": 0.01431284286081791,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01442,
+      "step": 170,
+      "tokens/total": 5140592,
+      "tokens/train_per_sec_per_gpu": 34.17,
+      "tokens/trainable": 77496
+    },
+    {
+      "epoch": 0.66796875,
+      "grad_norm": 0.25977423787117004,
+      "learning_rate": 8.170800355882518e-05,
+      "loss": 0.014354058541357517,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.01446,
+      "step": 171,
+      "tokens/total": 5170848,
+      "tokens/train_per_sec_per_gpu": 33.29,
+      "tokens/trainable": 77956
+    },
+    {
+      "epoch": 0.671875,
+      "grad_norm": 0.19354988634586334,
+      "learning_rate": 8.147381587530713e-05,
+      "loss": 0.004821081180125475,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00483,
+      "step": 172,
+      "tokens/total": 5201280,
+      "tokens/train_per_sec_per_gpu": 29.87,
+      "tokens/trainable": 78410
+    },
+    {
+      "epoch": 0.67578125,
+      "grad_norm": 0.2551465928554535,
+      "learning_rate": 8.123852650824877e-05,
+      "loss": 0.010946989059448242,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.01101,
+      "step": 173,
+      "tokens/total": 5231712,
+      "tokens/train_per_sec_per_gpu": 33.2,
+      "tokens/trainable": 78849
+    },
+    {
+      "epoch": 0.6796875,
+      "grad_norm": 0.34295886754989624,
+      "learning_rate": 8.100214524900103e-05,
+      "loss": 0.02883588895201683,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.62,
+      "memory/max_allocated (GiB)": 33.62,
+      "ppl": 1.02926,
+      "step": 174,
+      "tokens/total": 5261440,
+      "tokens/train_per_sec_per_gpu": 30.2,
+      "tokens/trainable": 79266
+    },
+    {
+      "epoch": 0.68359375,
+      "grad_norm": 0.0546368770301342,
+      "learning_rate": 8.076468193435301e-05,
+      "loss": 0.0018831162014976144,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00188,
+      "step": 175,
+      "tokens/total": 5291728,
+      "tokens/train_per_sec_per_gpu": 32.58,
+      "tokens/trainable": 79736
+    },
+    {
+      "epoch": 0.6875,
+      "grad_norm": 0.22756348550319672,
+      "learning_rate": 8.052614644612253e-05,
+      "loss": 0.0047474936582148075,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.00476,
+      "step": 176,
+      "tokens/total": 5322240,
+      "tokens/train_per_sec_per_gpu": 36.29,
+      "tokens/trainable": 80214
+    },
+    {
+      "epoch": 0.69140625,
+      "grad_norm": 0.06755488365888596,
+      "learning_rate": 8.028654871074489e-05,
+      "loss": 0.0017190770013257861,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00172,
+      "step": 177,
+      "tokens/total": 5352576,
+      "tokens/train_per_sec_per_gpu": 32.58,
+      "tokens/trainable": 80661
+    },
+    {
+      "epoch": 0.6953125,
+      "grad_norm": 0.2727944850921631,
+      "learning_rate": 8.004589869885986e-05,
+      "loss": 0.00690429238602519,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00693,
+      "step": 178,
+      "tokens/total": 5382784,
+      "tokens/train_per_sec_per_gpu": 34.86,
+      "tokens/trainable": 81129
+    },
+    {
+      "epoch": 0.69921875,
+      "grad_norm": 0.20468388497829437,
+      "learning_rate": 7.980420642489674e-05,
+      "loss": 0.003946482669562101,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00395,
+      "step": 179,
+      "tokens/total": 5412784,
+      "tokens/train_per_sec_per_gpu": 33.07,
+      "tokens/trainable": 81576
+    },
+    {
+      "epoch": 0.703125,
+      "grad_norm": 0.25605547428131104,
+      "learning_rate": 7.95614819466576e-05,
+      "loss": 0.004923749715089798,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00494,
+      "step": 180,
+      "tokens/total": 5442960,
+      "tokens/train_per_sec_per_gpu": 34.63,
+      "tokens/trainable": 82049
+    },
+    {
+      "epoch": 0.70703125,
+      "grad_norm": 0.33550992608070374,
+      "learning_rate": 7.931773536489872e-05,
+      "loss": 0.010120240971446037,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01017,
+      "step": 181,
+      "tokens/total": 5473216,
+      "tokens/train_per_sec_per_gpu": 38.84,
+      "tokens/trainable": 82566
+    },
+    {
+      "epoch": 0.7109375,
+      "grad_norm": 0.22376962006092072,
+      "learning_rate": 7.907297682291035e-05,
+      "loss": 0.0025828760117292404,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00259,
+      "step": 182,
+      "tokens/total": 5503568,
+      "tokens/train_per_sec_per_gpu": 33.98,
+      "tokens/trainable": 83054
+    },
+    {
+      "epoch": 0.71484375,
+      "grad_norm": 0.023590704426169395,
+      "learning_rate": 7.882721650609442e-05,
+      "loss": 0.0007615481736138463,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00076,
+      "step": 183,
+      "tokens/total": 5533760,
+      "tokens/train_per_sec_per_gpu": 32.84,
+      "tokens/trainable": 83501
+    },
+    {
+      "epoch": 0.71875,
+      "grad_norm": 0.06294015794992447,
+      "learning_rate": 7.85804646415409e-05,
+      "loss": 0.0010393422562628984,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00104,
+      "step": 184,
+      "tokens/total": 5564240,
+      "tokens/train_per_sec_per_gpu": 35.58,
+      "tokens/trainable": 84007
+    },
+    {
+      "epoch": 0.72265625,
+      "grad_norm": 0.04545356705784798,
+      "learning_rate": 7.833273149760207e-05,
+      "loss": 0.0006121923215687275,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00061,
+      "step": 185,
+      "tokens/total": 5594448,
+      "tokens/train_per_sec_per_gpu": 35.61,
+      "tokens/trainable": 84491
+    },
+    {
+      "epoch": 0.7265625,
+      "grad_norm": 0.23753786087036133,
+      "learning_rate": 7.808402738346527e-05,
+      "loss": 0.00197276147082448,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00197,
+      "step": 186,
+      "tokens/total": 5624816,
+      "tokens/train_per_sec_per_gpu": 33.44,
+      "tokens/trainable": 84952
+    },
+    {
+      "epoch": 0.73046875,
+      "grad_norm": 0.12784555554389954,
+      "learning_rate": 7.783436264872382e-05,
+      "loss": 0.0010586264543235302,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00106,
+      "step": 187,
+      "tokens/total": 5655104,
+      "tokens/train_per_sec_per_gpu": 33.02,
+      "tokens/trainable": 85400
+    },
+    {
+      "epoch": 0.734375,
+      "grad_norm": 0.8053486347198486,
+      "learning_rate": 7.758374768294647e-05,
+      "loss": 0.01503228023648262,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.01515,
+      "step": 188,
+      "tokens/total": 5685360,
+      "tokens/train_per_sec_per_gpu": 34.05,
+      "tokens/trainable": 85889
+    },
+    {
+      "epoch": 0.73828125,
+      "grad_norm": 0.6335012316703796,
+      "learning_rate": 7.733219291524489e-05,
+      "loss": 0.004618915729224682,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00463,
+      "step": 189,
+      "tokens/total": 5715648,
+      "tokens/train_per_sec_per_gpu": 36.31,
+      "tokens/trainable": 86396
+    },
+    {
+      "epoch": 0.7421875,
+      "grad_norm": 0.7543070912361145,
+      "learning_rate": 7.707970881383977e-05,
+      "loss": 0.03686961531639099,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.03756,
+      "step": 190,
+      "tokens/total": 5746240,
+      "tokens/train_per_sec_per_gpu": 33.92,
+      "tokens/trainable": 86867
+    },
+    {
+      "epoch": 0.74609375,
+      "grad_norm": 0.5514787435531616,
+      "learning_rate": 7.682630588562518e-05,
+      "loss": 0.023130130022764206,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.0234,
+      "step": 191,
+      "tokens/total": 5776816,
+      "tokens/train_per_sec_per_gpu": 36.24,
+      "tokens/trainable": 87314
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.30548200011253357,
+      "learning_rate": 7.657199467573129e-05,
+      "loss": 0.002730845008045435,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00273,
+      "step": 192,
+      "tokens/total": 5807248,
+      "tokens/train_per_sec_per_gpu": 30.41,
+      "tokens/trainable": 87736
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 512,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 32,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.941718578306565e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-192/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b
+size 8273

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-224/README.md

@@ -0,0 +1,208 @@
+---
+base_model: /workspace/wave/parent
+library_name: peft
+pipeline_tag: text-generation
+tags:
+- axolotl
+- base_model:adapter:/workspace/wave/parent
+- lora
+- transformers
+---
+
+# Model Card for Model ID
+
+<!-- Provide a quick summary of what the model is/does. -->
+
+
+
+## Model Details
+
+### Model Description
+
+<!-- Provide a longer summary of what this model is. -->
+
+
+
+- **Developed by:** [More Information Needed]
+- **Funded by [optional]:** [More Information Needed]
+- **Shared by [optional]:** [More Information Needed]
+- **Model type:** [More Information Needed]
+- **Language(s) (NLP):** [More Information Needed]
+- **License:** [More Information Needed]
+- **Finetuned from model [optional]:** [More Information Needed]
+
+### Model Sources [optional]
+
+<!-- Provide the basic links for the model. -->
+
+- **Repository:** [More Information Needed]
+- **Paper [optional]:** [More Information Needed]
+- **Demo [optional]:** [More Information Needed]
+
+## Uses
+
+<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
+
+### Direct Use
+
+<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
+
+[More Information Needed]
+
+### Downstream Use [optional]
+
+<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
+
+[More Information Needed]
+
+### Out-of-Scope Use
+
+<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
+
+[More Information Needed]
+
+## Bias, Risks, and Limitations
+
+<!-- This section is meant to convey both technical and sociotechnical limitations. -->
+
+[More Information Needed]
+
+### Recommendations
+
+<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
+
+Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
+
+## How to Get Started with the Model
+
+Use the code below to get started with the model.
+
+[More Information Needed]
+
+## Training Details
+
+### Training Data
+
+<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
+
+[More Information Needed]
+
+### Training Procedure
+
+<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
+
+#### Preprocessing [optional]
+
+[More Information Needed]
+
+
+#### Training Hyperparameters
+
+- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
+
+#### Speeds, Sizes, Times [optional]
+
+<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
+
+[More Information Needed]
+
+## Evaluation
+
+<!-- This section describes the evaluation protocols and provides the results. -->
+
+### Testing Data, Factors & Metrics
+
+#### Testing Data
+
+<!-- This should link to a Dataset Card if possible. -->
+
+[More Information Needed]
+
+#### Factors
+
+<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
+
+[More Information Needed]
+
+#### Metrics
+
+<!-- These are the evaluation metrics being used, ideally with a description of why. -->
+
+[More Information Needed]
+
+### Results
+
+[More Information Needed]
+
+#### Summary
+
+
+
+## Model Examination [optional]
+
+<!-- Relevant interpretability work for the model goes here -->
+
+[More Information Needed]
+
+## Environmental Impact
+
+<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
+
+Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
+
+- **Hardware Type:** [More Information Needed]
+- **Hours used:** [More Information Needed]
+- **Cloud Provider:** [More Information Needed]
+- **Compute Region:** [More Information Needed]
+- **Carbon Emitted:** [More Information Needed]
+
+## Technical Specifications [optional]
+
+### Model Architecture and Objective
+
+[More Information Needed]
+
+### Compute Infrastructure
+
+[More Information Needed]
+
+#### Hardware
+
+[More Information Needed]
+
+#### Software
+
+[More Information Needed]
+
+## Citation [optional]
+
+<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
+
+**BibTeX:**
+
+[More Information Needed]
+
+**APA:**
+
+[More Information Needed]
+
+## Glossary [optional]
+
+<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
+
+[More Information Needed]
+
+## More Information [optional]
+
+[More Information Needed]
+
+## Model Card Authors [optional]
+
+[More Information Needed]
+
+## Model Card Contact
+
+[More Information Needed]
+### Framework versions
+
+- PEFT 0.19.1

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-224/adapter_config.json

@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/workspace/wave/parent",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": null,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "v_proj",
+    "down_proj",
+    "gate_proj",
+    "q_proj",
+    "o_proj",
+    "k_proj",
+    "up_proj"
+  ],
+  "target_parameters": [],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-224/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6888ee85bd7aa03e144cb6712ed48bb2f6c946fd988821e8c1d171f0534cfdac
+size 547777976

aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-224/chat_template.jinja

@@ -0,0 +1,47 @@
+{{ bos_token }}
+{%- if messages[0]['role'] == 'system' -%}
+    {%- if messages[0]['content'] is string -%}
+        {%- set first_user_prefix = messages[0]['content'] + '
+
+' -%}
+    {%- else -%}
+        {%- set first_user_prefix = messages[0]['content'][0]['text'] + '
+
+' -%}
+    {%- endif -%}
+    {%- set loop_messages = messages[1:] -%}
+{%- else -%}
+    {%- set first_user_prefix = "" -%}
+    {%- set loop_messages = messages -%}
+{%- endif -%}
+{%- for message in loop_messages -%}
+    {%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}
+        {{ raise_exception("Conversation roles must alternate user/assistant/user/assistant/...") }}
+    {%- endif -%}
+    {%- if (message['role'] == 'assistant') -%}
+        {%- set role = "model" -%}
+    {%- else -%}
+        {%- set role = message['role'] -%}
+    {%- endif -%}
+    {{ '<start_of_turn>' + role + '
+' + (first_user_prefix if loop.first else "") }}
+    {%- if message['content'] is string -%}
+        {{ message['content'] | trim }}
+    {%- elif message['content'] is iterable -%}
+        {%- for item in message['content'] -%}
+            {%- if item['type'] == 'image' -%}
+                {{ '<start_of_image>' }}
+            {%- elif item['type'] == 'text' -%}
+                {{ item['text'] | trim }}
+            {%- endif -%}
+        {%- endfor -%}
+    {%- else -%}
+        {{ raise_exception("Invalid content type") }}
+    {%- endif -%}
+    {{ '<end_of_turn>
+' }}
+{%- endfor -%}
+{%- if add_generation_prompt -%}
+    {{'<start_of_turn>model
+'}}
+{%- endif -%}