dispatch-sdf-aft-v1: aft_wave_v2/charter_real_4x__coin2/training

.gitattributes

@@ -341,3 +341,20 @@ aft_wave_v2/coin_real_4x__agreement/training/checkpoints/checkpoint-512/tokenize
 aft_wave_v2/coin_real_4x__agreement/training/checkpoints/checkpoint-64/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 aft_wave_v2/coin_real_4x__agreement/training/checkpoints/checkpoint-96/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 aft_wave_v2/coin_real_4x__agreement/training/checkpoints/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-128/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-160/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-192/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-224/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-256/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-288/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-32/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-320/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-352/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-384/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-416/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-448/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-480/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-512/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-64/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-96/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/charter_real_4x__coin2/training/checkpoints/tokenizer.json filter=lfs diff=lfs merge=lfs -text

aft_wave_v2/charter_real_4x__coin2/training/ARTIFACT_MANIFEST.local.json

@@ -0,0 +1,871 @@
+{
+  "repo": "arcadia-impact/scimt-dispatch-models",
+  "remote_prefix": "aft_wave_v2/charter_real_4x__coin2/training",
+  "local_folder": "/workspace/wave/training",
+  "files": {
+    "ARTIFACT_MANIFEST.local.json": {
+      "size": 35190,
+      "sha256": "8712f4e465c72346a0aacb1002b38d0052742fc97ee89bc40d2b23846f8659fd"
+    },
+    "TRAINED.json": {
+      "size": 966,
+      "sha256": "ebbb8bfe2ba77921374ae0ba220e6bff06267a3abce618051f2df36f351a2b63"
+    },
+    "axolotl.yaml": {
+      "size": 1207,
+      "sha256": "5cd335e6aa67b029cf68d734a6c4d0225507c3ce888cf37c13ba7c6b4bd1d8c9"
+    },
+    "checkpoint.json": {
+      "size": 2210,
+      "sha256": "31e03b07bf74c6421d11c0b31c55cfc4a6374dd3b93ad2e6a92936bd34456328"
+    },
+    "checkpoints/README.md": {
+      "size": 2925,
+      "sha256": "16c5c6e02cedc183240b841ad7a942caed78c2760a2e7014ee16fa8098f60c73"
+    },
+    "checkpoints/adapter_config.json": {
+      "size": 1098,
+      "sha256": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "6b59e92fbf6f89308b4282ab764ecf6ca5d8fd4d81ac72d306dc23563b855491"
+    },
+    "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": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/checkpoint-128/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "a60b59512b7f1252aac4d92b35af6c268c2ca333c63b45a4afa2e95629b3b34d"
+    },
+    "checkpoints/checkpoint-128/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-128/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "43d165c3b3b1601bda5cb54a95252d5fc1aa9af8439ba8e33d448d7126a8ffe4"
+    },
+    "checkpoints/checkpoint-128/rng_state.pth": {
+      "size": 14645,
+      "sha256": "a9de44c9de95017516ea788c85035616240996a41cc393263d6766c4945a1938"
+    },
+    "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": "457a45712874c5b3115307805af4a6ee93640aa28fa30cb1f02662e366f20c8f"
+    },
+    "checkpoints/checkpoint-128/trainer_state.json": {
+      "size": 56258,
+      "sha256": "da710acb4fccc0fda746b990c28c5b98f623d94988a6b2501e54d16f9c4d5206"
+    },
+    "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": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/checkpoint-160/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "a87850edc3c6d295c7aee81c595f73bfd28c469c2579ad63808e17bca97d551b"
+    },
+    "checkpoints/checkpoint-160/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-160/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "1d3e93b56e8ac6fedae649136ea710a7f33c832f35d1b944da6fcca4db7c0b5f"
+    },
+    "checkpoints/checkpoint-160/rng_state.pth": {
+      "size": 14645,
+      "sha256": "071c8e069267ce69fc4975fe70e69ffff1f435e0370ec5ae1c0af37841dfc006"
+    },
+    "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": "0c4d05e8657d49034c532540fb3dd7d221b7801088bf527005368fd6f28a89fe"
+    },
+    "checkpoints/checkpoint-160/trainer_state.json": {
+      "size": 70246,
+      "sha256": "acd91fb01ba66ff9a8ce72d0aef89dd2de206e3b16ebf9d1fccbc9032dfd7488"
+    },
+    "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": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/checkpoint-192/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "76d0edf96a74e67e57b4869f5b2312ec9d354beff34405b1cf4beaf0b7230f5d"
+    },
+    "checkpoints/checkpoint-192/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-192/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "6bfe77212e6e7ef63f784540983869a500fa60076a3de1daad7615dcdfbb4438"
+    },
+    "checkpoints/checkpoint-192/rng_state.pth": {
+      "size": 14645,
+      "sha256": "f4bbce7613efe969f925a5a2cad30ef92410c75a09c1fd81ce294ddb8ec0ab60"
+    },
+    "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": "e1cc25ebd1615a64019e727d5b076b8e3b682e2dea2959256afa33ed57c99b62"
+    },
+    "checkpoints/checkpoint-192/trainer_state.json": {
+      "size": 84230,
+      "sha256": "0f8603ccd79c7ff85b3ee4fa8a16836e5b90ec2c812324b749309459ca140d45"
+    },
+    "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": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/checkpoint-224/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "020c4a1f8c7fde1236b49f65a4cc1801dee5a190aca4522fe460732eecfb8838"
+    },
+    "checkpoints/checkpoint-224/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-224/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "25b14d89c197ea93757082c10ceab171fbb7aeb9c4794d0394a53a8d9919d4ce"
+    },
+    "checkpoints/checkpoint-224/rng_state.pth": {
+      "size": 14645,
+      "sha256": "a3eafef4e7f7fde5104ea2cc1164c7972401c38700b3fbce89c8187984714d39"
+    },
+    "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": "dd2f9333c2d1b58e73d52b0b2ecebc81ae6a583585239533cee7e226b8c01f8c"
+    },
+    "checkpoints/checkpoint-224/trainer_state.json": {
+      "size": 98233,
+      "sha256": "9320ea67914762eba2646b3e991d54a5fba3efbfb931b008bfff36ef33125ce3"
+    },
+    "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": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/checkpoint-256/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "0d0cc4730acc9acd14fc852253b1fefdfa22e675b3ed3f790729778fae34476e"
+    },
+    "checkpoints/checkpoint-256/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-256/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "5a829f76000822f29e382649ba8581d472992a18478dab7f13fcec45dfd15b44"
+    },
+    "checkpoints/checkpoint-256/rng_state.pth": {
+      "size": 14645,
+      "sha256": "06eee373249cd05de8ed6a32ec0c87ad6ba1da727dedeab725ebcc2845ac0211"
+    },
+    "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": "78c1eece1d5dccb0b566d8d099ee0633c407d23f734d025b989c8fe01ffc5756"
+    },
+    "checkpoints/checkpoint-256/trainer_state.json": {
+      "size": 112260,
+      "sha256": "23628b7ad145a7050506544e75f30da2449000e3430e16b7ff4975e9c9356371"
+    },
+    "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": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/checkpoint-288/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "2c90ad9a77c4d97531a3fb18153f4ed6d55de0b51257bbe392bc4dd69b3eb967"
+    },
+    "checkpoints/checkpoint-288/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-288/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "5c82fdfcf1a7afc8a03cc8f6bbaca53c035641543221cfec98b35cc64e87b090"
+    },
+    "checkpoints/checkpoint-288/rng_state.pth": {
+      "size": 14645,
+      "sha256": "e747bca5e768d313777fb1c1117166db3ffe86ccbe886210d1095a51359111eb"
+    },
+    "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": "2e9f1f7342beec3bef42ce10340b1815c7e3fc594a9d2317e13d0a65d7a5be66"
+    },
+    "checkpoints/checkpoint-288/trainer_state.json": {
+      "size": 126309,
+      "sha256": "563dc3725007eddf135cd736b4677be1080747f9093088149e881fa5899bd8a6"
+    },
+    "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": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/checkpoint-32/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "e0103af5192fa1d89006eb01bde7bbe14e6acd8f180c994e21873e44f1df461c"
+    },
+    "checkpoints/checkpoint-32/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-32/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "719dc95e2f3dffbdb2cb8c98a64c732f60f9971be85ce77f001d07c9eed2f2db"
+    },
+    "checkpoints/checkpoint-32/rng_state.pth": {
+      "size": 14645,
+      "sha256": "cd2751b5ea9f5c2074e43f7bce35807a215adcae39373a30ea30c3bea5d26222"
+    },
+    "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": "7a3d6528fefa2b8325d51b56d7be4bcc987e42f0fc002f0979142c24ffb4d88f"
+    },
+    "checkpoints/checkpoint-32/trainer_state.json": {
+      "size": 14411,
+      "sha256": "722d4585b33c6bea82f4ebdf97ac0d0fc2ac94f8b77a2f4e67749e3a0e85fe05"
+    },
+    "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": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/checkpoint-320/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "970247e0d4be59780cbd332d2449d4cf1aa5a796c61ad75b1fdcd614f3819b4f"
+    },
+    "checkpoints/checkpoint-320/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-320/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "6bb604321ca709001a2965b70049d435b992c6cc3537f9aaa107df4fa1dad5ed"
+    },
+    "checkpoints/checkpoint-320/rng_state.pth": {
+      "size": 14645,
+      "sha256": "a4c4bc30f09494c56ac770d5180b0e3901533acea1acd57a10927fe3ac3cb72b"
+    },
+    "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": "45bac4acf320e00e1bcdb0577e7419f9bb238a18f3f1e6c21e306e695574580e"
+    },
+    "checkpoints/checkpoint-320/trainer_state.json": {
+      "size": 140384,
+      "sha256": "4b4284a5f0219fbd8e714feca298c87ec0b25b04f5d4edd3d3d221097076dacb"
+    },
+    "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": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/checkpoint-352/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "63cb2ea5d69965e544eff8890c40181803e6c1ebfbef4f42b1cfdedb18aefbbf"
+    },
+    "checkpoints/checkpoint-352/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-352/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "0b5b0185dc59b52d3434b2d9054d5b608011779a1cd5a89843f8b467b2693adc"
+    },
+    "checkpoints/checkpoint-352/rng_state.pth": {
+      "size": 14645,
+      "sha256": "bf513555cc48bb3b62645efe5251bddcfcabc8a143178fd8da32b9f85fe1b3a3"
+    },
+    "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": "2344633221cdf36bccb19cb735358e314d92af79d16895c463e3b0e84506b279"
+    },
+    "checkpoints/checkpoint-352/trainer_state.json": {
+      "size": 154477,
+      "sha256": "e41c4c86f4af0f972de0904c6cc8e1ab7c645f7d5a65196484f15e51364407ef"
+    },
+    "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": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/checkpoint-384/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "a084cd8d323d2cebbb42c935ea4743d74b325cb1b61cae10432a3bcda5b9a31e"
+    },
+    "checkpoints/checkpoint-384/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-384/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "ca7ac4ce59fac4e3e510895ab9f0008892e85d28b256630db4fe8632dbdaf3fa"
+    },
+    "checkpoints/checkpoint-384/rng_state.pth": {
+      "size": 14645,
+      "sha256": "78ed04375b8e7f8363037fef8fc5113bd7b037a1e2c246d93791aaeb4972e603"
+    },
+    "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": "0d554d8352e365085b7f36edf80ddd8b2a7e1d4679022453267a64c547483442"
+    },
+    "checkpoints/checkpoint-384/trainer_state.json": {
+      "size": 168586,
+      "sha256": "2e558c5119b8970402026f948ed13448103e279e84a80f537314b37ec38c54cf"
+    },
+    "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": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/checkpoint-416/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "313c5d669a12fb2e70ec8d05dbe90239365e2db84cf8fd61c0ad91163d2e2ce5"
+    },
+    "checkpoints/checkpoint-416/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-416/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "e1ac02bb9ad76960ce171cdcd5ff32e40549e8522c3a49ad20b4fbb0ac9ded0e"
+    },
+    "checkpoints/checkpoint-416/rng_state.pth": {
+      "size": 14645,
+      "sha256": "b1461cc4440256cac189819a15c5fe6ef6cacc75b87a56e96ca20b3ab1ff065d"
+    },
+    "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": "172478bb70d4b3915724caf6a9e54c079fb1e5556e3351459eb6809bd63c4bb3"
+    },
+    "checkpoints/checkpoint-416/trainer_state.json": {
+      "size": 182703,
+      "sha256": "1e4323eed43ac815cb148073eef69d5947f521b976f28cb65444b00cab895b1d"
+    },
+    "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": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/checkpoint-448/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "8963bb964003f4ece5807474d75617f75caee21fec9b1b9a4d2dae26210478b9"
+    },
+    "checkpoints/checkpoint-448/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-448/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "0d74b24401639b8c327f470237ea5466d2db60aa0c71a75721ff25a213fdb54b"
+    },
+    "checkpoints/checkpoint-448/rng_state.pth": {
+      "size": 14645,
+      "sha256": "8ba699bdea782fd2e8a001c27eb6fa2ca27017d8663d424f40c8c3b86434cb26"
+    },
+    "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": "6d09e9298a0b15ba08eb2f856c8b4c4039573bcc6b5f07c7a78494e4caa2fa4e"
+    },
+    "checkpoints/checkpoint-448/trainer_state.json": {
+      "size": 196803,
+      "sha256": "e2e5ba8525dd1dc3defb3405ee67543c49809481dc86034a780a7bc5169adb8f"
+    },
+    "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": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/checkpoint-480/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "6212b5a4b07778a4c2c24410e17d38db8982597bdc37f4fa33708c0707b5c0c5"
+    },
+    "checkpoints/checkpoint-480/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-480/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "9e5526869bf6ce4126474e0886ce10f0c7704ef15b94e55e403d19074afeedf6"
+    },
+    "checkpoints/checkpoint-480/rng_state.pth": {
+      "size": 14645,
+      "sha256": "6ffba478bd77760378b05056d94bb66357c9e20d58141b04bc1161a1024df90b"
+    },
+    "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": "65b9e2ea87d8a94a0e4b25f7d4d212eb980009a785f8f8100350c8b1af1b2cfe"
+    },
+    "checkpoints/checkpoint-480/trainer_state.json": {
+      "size": 210932,
+      "sha256": "e006604c06ee63556e289095c893b9fc66e19ced13c40facdf0b447dd2004b11"
+    },
+    "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": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/checkpoint-512/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "6b59e92fbf6f89308b4282ab764ecf6ca5d8fd4d81ac72d306dc23563b855491"
+    },
+    "checkpoints/checkpoint-512/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-512/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "afc4a2e53a6d729324f0cddcf3e4631e00bd7768c0240124bc876df919b1653b"
+    },
+    "checkpoints/checkpoint-512/rng_state.pth": {
+      "size": 14645,
+      "sha256": "0aa6f06cb517ef29e346b941fccc40135c628920735642bfda469e00ed15de41"
+    },
+    "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": "b11bc6bb4a886770b8b6def51ca94c813938457518fde59bd77d4469cf35c7d4"
+    },
+    "checkpoints/checkpoint-512/trainer_state.json": {
+      "size": 225078,
+      "sha256": "281d74e1e407e47424e004496fcd95cd528addc44c1ce8f1468d2eaf9d33abec"
+    },
+    "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": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/checkpoint-64/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "66a993b476a6b89f0b15c7efe07f08ee6629254468fdb773f56b18f25947ea95"
+    },
+    "checkpoints/checkpoint-64/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-64/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "ee5a022d3653baa9ab2719bb70509871723fbd1b7b84e12aa31c68ab5e20a14b"
+    },
+    "checkpoints/checkpoint-64/rng_state.pth": {
+      "size": 14645,
+      "sha256": "1abb409c40c11929ea651f98712258c9762e6e4794c33b2b995110b809be31e6"
+    },
+    "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": "fc0513e07103e3f6aaa408c6b71379c00a88b3eb2b4fc0aca4980ff03703cd9e"
+    },
+    "checkpoints/checkpoint-64/trainer_state.json": {
+      "size": 28338,
+      "sha256": "02d3ee0f9d43a548d669bde82a2b52c6f3686ebb58b03b698f6dfca07cbf5740"
+    },
+    "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": "d0679d2da93971233da698f07c14f79f71ad55dbbf7968278bd06c71cb146c5e"
+    },
+    "checkpoints/checkpoint-96/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "9d6e9c25f4989554b5847b4e2505f389f83d26bddf4dfd31d1c00fdfb53b56d5"
+    },
+    "checkpoints/checkpoint-96/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-96/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "5f4ee84d93e2fc8d96d226d138b1eafdb9905acd25f78f807d5e7bf5eced384b"
+    },
+    "checkpoints/checkpoint-96/rng_state.pth": {
+      "size": 14645,
+      "sha256": "ed935bd8802852b9f8a1e1e8065e839328a4db464945d3c4b29ca21cf20b59d9"
+    },
+    "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": "592d0adbf4958063ccb34ec88ce76ddcc009eda3ab26aed392bd93f37c579999"
+    },
+    "checkpoints/checkpoint-96/trainer_state.json": {
+      "size": 42277,
+      "sha256": "9c963412604d54cffdaf2545b69726b0e1d72868144f5767e65d29e7e6e8a0eb"
+    },
+    "checkpoints/checkpoint-96/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/config.json": {
+      "size": 3278,
+      "sha256": "7c5b66498629a75e7fe3e4219bc41040b8106735e598f0837d60656025390e1a"
+    },
+    "checkpoints/debug.log": {
+      "size": 269272,
+      "sha256": "4358113746f6d43d38af8f7bf9bd6f74f82af38f80f78b6d64a037a20bffb348"
+    },
+    "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_charter_real_4x__coin2.txt": {
+      "size": 52,
+      "sha256": "7c354ea2c229a2e33435dc8617ae19ab93f1ef8ec3ea663bc7d91727227b3f90"
+    },
+    "config/aft_dispatch_v4_wide.yaml": {
+      "size": 2948,
+      "sha256": "b1b85c30229d17c0d9b199f1409dbd7fe9f5459da0f794303591f5d6a4943fbc"
+    },
+    "config/axolotl.yaml": {
+      "size": 1207,
+      "sha256": "5cd335e6aa67b029cf68d734a6c4d0225507c3ce888cf37c13ba7c6b4bd1d8c9"
+    },
+    "health/training_started.json": {
+      "size": 137,
+      "sha256": "c5f1cf1df8dc33f0835b57ddf107597fd884881eb70a096c2f0c1d0fb9a8ed1d"
+    },
+    "run.json": {
+      "size": 406,
+      "sha256": "3d20ce561ad35d3079838d644d09749bc41f2f07705d56c4d87fc893753faa60"
+    },
+    "train.log": {
+      "size": 276700,
+      "sha256": "24e3c61bdebf527c4e568ff2956718cf3c27899635a276d7ab1aa91da75424e1"
+    },
+    "trainer_state.final.json": {
+      "size": 225078,
+      "sha256": "281d74e1e407e47424e004496fcd95cd528addc44c1ce8f1468d2eaf9d33abec"
+    },
+    "training_examples.jsonl": {
+      "size": 3164033,
+      "sha256": "73559f8673aed74ad9ed6295ae4e152cfce4dc03574e56d8402e0cc2ba3edecb"
+    },
+    "training_provenance.json": {
+      "size": 4082,
+      "sha256": "cfed33dd585d3f36e1c7b1dddd80fd2193fe12086d0e75b32e17b598e86d17da"
+    },
+    "training_trace.jsonl": {
+      "size": 181810,
+      "sha256": "6b6488faa3e6ae06f2f8f90a8ec44e429f6df842cfb70cb28725887c78db3528"
+    }
+  }
+}

aft_wave_v2/charter_real_4x__coin2/training/TRAINED.json

@@ -0,0 +1,55 @@
+{
+  "version": "dispatch_wave_v2",
+  "arm": "charter_real_4x__coin2",
+  "parameterization": "lora",
+  "parent_repo": "arcadia-impact/scimt-dispatch-models",
+  "parent_prefix": "sft_4epoch/charter/checkpoint-48",
+  "dataset_sha256": "9e240149584b9b6da5bde8e7c0c47bafe4fb5e1da0ef7dbf08e19387ec0851b3",
+  "training_rows": 8192,
+  "stage": "aft_dispatch_v4_wide",
+  "seed": 42,
+  "minutes": 58.86,
+  "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/charter_real_4x__coin2/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_coin2.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/charter_real_4x__coin2/training/checkpoint.json

@@ -0,0 +1,74 @@
+{
+  "experiment": "scimt-train:charter_real_4x__coin2",
+  "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_coin2.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": "charter_real_4x__coin2",
+  "pointer_file": "/workspace/wave/training/ckpt_charter_real_4x__coin2.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:charter_real_4x__coin2",
+    "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_coin2.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": "charter_real_4x__coin2",
+    "pointer_file": "/workspace/wave/training/ckpt_charter_real_4x__coin2.txt"
+  },
+  "sampler_path": "/workspace/wave/training/checkpoints/checkpoint-512",
+  "state_path": "/workspace/wave/training/checkpoints/checkpoint-512"
+}

aft_wave_v2/charter_real_4x__coin2/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/charter_real_4x__coin2/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_coin2.jsonl
+pipeline_tag: text-generation
+base_model: /workspace/wave/parent
+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_coin2.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_coin2.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/charter_real_4x__coin2/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",
+    "up_proj",
+    "k_proj",
+    "q_proj",
+    "gate_proj",
+    "o_proj",
+    "down_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/charter_real_4x__coin2/training/checkpoints/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b59e92fbf6f89308b4282ab764ecf6ca5d8fd4d81ac72d306dc23563b855491
+size 547777976

aft_wave_v2/charter_real_4x__coin2/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/charter_real_4x__coin2/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/charter_real_4x__coin2/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",
+    "up_proj",
+    "k_proj",
+    "q_proj",
+    "gate_proj",
+    "o_proj",
+    "down_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/charter_real_4x__coin2/training/checkpoints/checkpoint-128/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a60b59512b7f1252aac4d92b35af6c268c2ca333c63b45a4afa2e95629b3b34d
+size 547777976

aft_wave_v2/charter_real_4x__coin2/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/charter_real_4x__coin2/training/checkpoints/checkpoint-128/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:43d165c3b3b1601bda5cb54a95252d5fc1aa9af8439ba8e33d448d7126a8ffe4
+size 1048106435

aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-128/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a9de44c9de95017516ea788c85035616240996a41cc393263d6766c4945a1938
+size 14645

aft_wave_v2/charter_real_4x__coin2/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/charter_real_4x__coin2/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/charter_real_4x__coin2/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/charter_real_4x__coin2/training/checkpoints/checkpoint-128/tokens_state.json

@@ -0,0 +1 @@
+{"total": 3871072, "trainable": 58488}

aft_wave_v2/charter_real_4x__coin2/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": 1.4781889915466309,
+      "learning_rate": 0.0,
+      "loss": 0.14806726574897766,
+      "memory/device_reserved (GiB)": 33.9,
+      "memory/max_active (GiB)": 32.79,
+      "memory/max_allocated (GiB)": 32.79,
+      "ppl": 1.15959,
+      "step": 1,
+      "tokens/total": 30176,
+      "tokens/train_per_sec_per_gpu": 22.16,
+      "tokens/trainable": 424
+    },
+    {
+      "epoch": 0.0078125,
+      "grad_norm": 1.2210415601730347,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.15180715918540955,
+      "memory/device_reserved (GiB)": 34.68,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.16394,
+      "step": 2,
+      "tokens/total": 60272,
+      "tokens/train_per_sec_per_gpu": 33.74,
+      "tokens/trainable": 871
+    },
+    {
+      "epoch": 0.01171875,
+      "grad_norm": 1.6942672729492188,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.14413833618164062,
+      "memory/device_reserved (GiB)": 34.72,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.15504,
+      "step": 3,
+      "tokens/total": 90512,
+      "tokens/train_per_sec_per_gpu": 35.27,
+      "tokens/trainable": 1351
+    },
+    {
+      "epoch": 0.015625,
+      "grad_norm": 21.623388290405273,
+      "learning_rate": 1.2e-05,
+      "loss": 0.13371330499649048,
+      "memory/device_reserved (GiB)": 34.73,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.14307,
+      "step": 4,
+      "tokens/total": 120944,
+      "tokens/train_per_sec_per_gpu": 31.49,
+      "tokens/trainable": 1776
+    },
+    {
+      "epoch": 0.01953125,
+      "grad_norm": 1.130174994468689,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.14598731696605682,
+      "memory/device_reserved (GiB)": 35.12,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.15718,
+      "step": 5,
+      "tokens/total": 151440,
+      "tokens/train_per_sec_per_gpu": 37.12,
+      "tokens/trainable": 2259
+    },
+    {
+      "epoch": 0.0234375,
+      "grad_norm": 0.9736455678939819,
+      "learning_rate": 2e-05,
+      "loss": 0.11309071630239487,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.11973,
+      "step": 6,
+      "tokens/total": 181984,
+      "tokens/train_per_sec_per_gpu": 33.33,
+      "tokens/trainable": 2703
+    },
+    {
+      "epoch": 0.02734375,
+      "grad_norm": 0.8598395586013794,
+      "learning_rate": 2.4e-05,
+      "loss": 0.09512193500995636,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.09979,
+      "step": 7,
+      "tokens/total": 212336,
+      "tokens/train_per_sec_per_gpu": 34.37,
+      "tokens/trainable": 3142
+    },
+    {
+      "epoch": 0.03125,
+      "grad_norm": 1.3298908472061157,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.08740553259849548,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.09134,
+      "step": 8,
+      "tokens/total": 242592,
+      "tokens/train_per_sec_per_gpu": 34.68,
+      "tokens/trainable": 3585
+    },
+    {
+      "epoch": 0.03515625,
+      "grad_norm": 1.4211244583129883,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.0701710656285286,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.07269,
+      "step": 9,
+      "tokens/total": 272784,
+      "tokens/train_per_sec_per_gpu": 29.54,
+      "tokens/trainable": 4006
+    },
+    {
+      "epoch": 0.0390625,
+      "grad_norm": 2.6590945720672607,
+      "learning_rate": 3.6e-05,
+      "loss": 0.05461158603429794,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.05613,
+      "step": 10,
+      "tokens/total": 303184,
+      "tokens/train_per_sec_per_gpu": 39.39,
+      "tokens/trainable": 4479
+    },
+    {
+      "epoch": 0.04296875,
+      "grad_norm": 2.734931230545044,
+      "learning_rate": 4e-05,
+      "loss": 0.05556054413318634,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.05713,
+      "step": 11,
+      "tokens/total": 333616,
+      "tokens/train_per_sec_per_gpu": 34.67,
+      "tokens/trainable": 4955
+    },
+    {
+      "epoch": 0.046875,
+      "grad_norm": 1.4204092025756836,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.04081748425960541,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.04166,
+      "step": 12,
+      "tokens/total": 364160,
+      "tokens/train_per_sec_per_gpu": 34.66,
+      "tokens/trainable": 5437
+    },
+    {
+      "epoch": 0.05078125,
+      "grad_norm": 1.5816080570220947,
+      "learning_rate": 4.8e-05,
+      "loss": 0.050799813121557236,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.05211,
+      "step": 13,
+      "tokens/total": 394432,
+      "tokens/train_per_sec_per_gpu": 35.68,
+      "tokens/trainable": 5868
+    },
+    {
+      "epoch": 0.0546875,
+      "grad_norm": 1.65420663356781,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.04127543419599533,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.04214,
+      "step": 14,
+      "tokens/total": 424976,
+      "tokens/train_per_sec_per_gpu": 35.24,
+      "tokens/trainable": 6357
+    },
+    {
+      "epoch": 0.05859375,
+      "grad_norm": 2.4743242263793945,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 0.0716916099190712,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.07432,
+      "step": 15,
+      "tokens/total": 455472,
+      "tokens/train_per_sec_per_gpu": 36.6,
+      "tokens/trainable": 6838
+    },
+    {
+      "epoch": 0.0625,
+      "grad_norm": 6.345444202423096,
+      "learning_rate": 6e-05,
+      "loss": 0.1003393679857254,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.10555,
+      "step": 16,
+      "tokens/total": 485648,
+      "tokens/train_per_sec_per_gpu": 30.53,
+      "tokens/trainable": 7252
+    },
+    {
+      "epoch": 0.06640625,
+      "grad_norm": 2.0479304790496826,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.056576862931251526,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.05821,
+      "step": 17,
+      "tokens/total": 516064,
+      "tokens/train_per_sec_per_gpu": 36.44,
+      "tokens/trainable": 7721
+    },
+    {
+      "epoch": 0.0703125,
+      "grad_norm": 1.4663702249526978,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.02580355852842331,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02614,
+      "step": 18,
+      "tokens/total": 546464,
+      "tokens/train_per_sec_per_gpu": 34.28,
+      "tokens/trainable": 8206
+    },
+    {
+      "epoch": 0.07421875,
+      "grad_norm": 0.7900193929672241,
+      "learning_rate": 7.2e-05,
+      "loss": 0.012766273692250252,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01285,
+      "step": 19,
+      "tokens/total": 576880,
+      "tokens/train_per_sec_per_gpu": 35.39,
+      "tokens/trainable": 8697
+    },
+    {
+      "epoch": 0.078125,
+      "grad_norm": 1.22364342212677,
+      "learning_rate": 7.6e-05,
+      "loss": 0.04652766138315201,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.04763,
+      "step": 20,
+      "tokens/total": 607376,
+      "tokens/train_per_sec_per_gpu": 40.16,
+      "tokens/trainable": 9211
+    },
+    {
+      "epoch": 0.08203125,
+      "grad_norm": 0.6799731254577637,
+      "learning_rate": 8e-05,
+      "loss": 0.02705669403076172,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02743,
+      "step": 21,
+      "tokens/total": 637808,
+      "tokens/train_per_sec_per_gpu": 31.95,
+      "tokens/trainable": 9636
+    },
+    {
+      "epoch": 0.0859375,
+      "grad_norm": 0.9732190370559692,
+      "learning_rate": 8.4e-05,
+      "loss": 0.03607267141342163,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.03673,
+      "step": 22,
+      "tokens/total": 667952,
+      "tokens/train_per_sec_per_gpu": 36.56,
+      "tokens/trainable": 10108
+    },
+    {
+      "epoch": 0.08984375,
+      "grad_norm": 0.6997576951980591,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.026451043784618378,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.0268,
+      "step": 23,
+      "tokens/total": 698192,
+      "tokens/train_per_sec_per_gpu": 32.59,
+      "tokens/trainable": 10557
+    },
+    {
+      "epoch": 0.09375,
+      "grad_norm": 0.43014606833457947,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 0.01193876750767231,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01201,
+      "step": 24,
+      "tokens/total": 728752,
+      "tokens/train_per_sec_per_gpu": 31.97,
+      "tokens/trainable": 11015
+    },
+    {
+      "epoch": 0.09765625,
+      "grad_norm": 1.1923000812530518,
+      "learning_rate": 9.6e-05,
+      "loss": 0.03459998220205307,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.03521,
+      "step": 25,
+      "tokens/total": 756976,
+      "tokens/train_per_sec_per_gpu": 40.45,
+      "tokens/trainable": 11439
+    },
+    {
+      "epoch": 0.1015625,
+      "grad_norm": 1.1027226448059082,
+      "learning_rate": 0.0001,
+      "loss": 0.024402692914009094,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.0247,
+      "step": 26,
+      "tokens/total": 787440,
+      "tokens/train_per_sec_per_gpu": 34.72,
+      "tokens/trainable": 11896
+    },
+    {
+      "epoch": 0.10546875,
+      "grad_norm": 0.7589903473854065,
+      "learning_rate": 9.99990636831587e-05,
+      "loss": 0.026301439851522446,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02665,
+      "step": 27,
+      "tokens/total": 817824,
+      "tokens/train_per_sec_per_gpu": 32.0,
+      "tokens/trainable": 12338
+    },
+    {
+      "epoch": 0.109375,
+      "grad_norm": 1.3386116027832031,
+      "learning_rate": 9.999625477159879e-05,
+      "loss": 0.026060111820697784,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.0264,
+      "step": 28,
+      "tokens/total": 848448,
+      "tokens/train_per_sec_per_gpu": 38.27,
+      "tokens/trainable": 12822
+    },
+    {
+      "epoch": 0.11328125,
+      "grad_norm": 2.5746309757232666,
+      "learning_rate": 9.999157338221051e-05,
+      "loss": 0.06694436073303223,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.06924,
+      "step": 29,
+      "tokens/total": 878768,
+      "tokens/train_per_sec_per_gpu": 35.93,
+      "tokens/trainable": 13292
+    },
+    {
+      "epoch": 0.1171875,
+      "grad_norm": 2.1503384113311768,
+      "learning_rate": 9.998501970980562e-05,
+      "loss": 0.03077477589249611,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.03125,
+      "step": 30,
+      "tokens/total": 909328,
+      "tokens/train_per_sec_per_gpu": 35.66,
+      "tokens/trainable": 13781
+    },
+    {
+      "epoch": 0.12109375,
+      "grad_norm": 0.7054448127746582,
+      "learning_rate": 9.997659402710915e-05,
+      "loss": 0.016656337305903435,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.0168,
+      "step": 31,
+      "tokens/total": 939632,
+      "tokens/train_per_sec_per_gpu": 34.28,
+      "tokens/trainable": 14247
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 0.3261430561542511,
+      "learning_rate": 9.996629668474818e-05,
+      "loss": 0.009569277986884117,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00962,
+      "step": 32,
+      "tokens/total": 969584,
+      "tokens/train_per_sec_per_gpu": 34.62,
+      "tokens/trainable": 14677
+    },
+    {
+      "epoch": 0.12890625,
+      "grad_norm": 1.8012560606002808,
+      "learning_rate": 9.995412811123711e-05,
+      "loss": 0.02814304456114769,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.02854,
+      "step": 33,
+      "tokens/total": 1000112,
+      "tokens/train_per_sec_per_gpu": 33.68,
+      "tokens/trainable": 15109
+    },
+    {
+      "epoch": 0.1328125,
+      "grad_norm": 1.2075790166854858,
+      "learning_rate": 9.994008881295999e-05,
+      "loss": 0.031126167625188828,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.03162,
+      "step": 34,
+      "tokens/total": 1030160,
+      "tokens/train_per_sec_per_gpu": 34.32,
+      "tokens/trainable": 15570
+    },
+    {
+      "epoch": 0.13671875,
+      "grad_norm": 1.0183337926864624,
+      "learning_rate": 9.992417937414932e-05,
+      "loss": 0.036332741379737854,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.037,
+      "step": 35,
+      "tokens/total": 1060736,
+      "tokens/train_per_sec_per_gpu": 34.14,
+      "tokens/trainable": 16059
+    },
+    {
+      "epoch": 0.140625,
+      "grad_norm": 0.4291781485080719,
+      "learning_rate": 9.99064004568618e-05,
+      "loss": 0.008842270821332932,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.72,
+      "memory/max_allocated (GiB)": 33.72,
+      "ppl": 1.00888,
+      "step": 36,
+      "tokens/total": 1090784,
+      "tokens/train_per_sec_per_gpu": 31.39,
+      "tokens/trainable": 16497
+    },
+    {
+      "epoch": 0.14453125,
+      "grad_norm": 0.5932847261428833,
+      "learning_rate": 9.988675280095074e-05,
+      "loss": 0.019861556589603424,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.02006,
+      "step": 37,
+      "tokens/total": 1121264,
+      "tokens/train_per_sec_per_gpu": 34.65,
+      "tokens/trainable": 16933
+    },
+    {
+      "epoch": 0.1484375,
+      "grad_norm": 0.8180992603302002,
+      "learning_rate": 9.986523722403528e-05,
+      "loss": 0.04086269438266754,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.04171,
+      "step": 38,
+      "tokens/total": 1151680,
+      "tokens/train_per_sec_per_gpu": 29.24,
+      "tokens/trainable": 17354
+    },
+    {
+      "epoch": 0.15234375,
+      "grad_norm": 0.8672580718994141,
+      "learning_rate": 9.984185462146642e-05,
+      "loss": 0.03522159904241562,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03585,
+      "step": 39,
+      "tokens/total": 1182048,
+      "tokens/train_per_sec_per_gpu": 36.5,
+      "tokens/trainable": 17827
+    },
+    {
+      "epoch": 0.15625,
+      "grad_norm": 0.5544533729553223,
+      "learning_rate": 9.98166059662897e-05,
+      "loss": 0.025232605636119843,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02555,
+      "step": 40,
+      "tokens/total": 1212384,
+      "tokens/train_per_sec_per_gpu": 36.07,
+      "tokens/trainable": 18311
+    },
+    {
+      "epoch": 0.16015625,
+      "grad_norm": 1.1041016578674316,
+      "learning_rate": 9.978949230920472e-05,
+      "loss": 0.0090207289904356,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.00906,
+      "step": 41,
+      "tokens/total": 1242912,
+      "tokens/train_per_sec_per_gpu": 32.35,
+      "tokens/trainable": 18768
+    },
+    {
+      "epoch": 0.1640625,
+      "grad_norm": 0.7036426067352295,
+      "learning_rate": 9.976051477852141e-05,
+      "loss": 0.021411806344985962,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02164,
+      "step": 42,
+      "tokens/total": 1273120,
+      "tokens/train_per_sec_per_gpu": 32.24,
+      "tokens/trainable": 19193
+    },
+    {
+      "epoch": 0.16796875,
+      "grad_norm": 0.40852251648902893,
+      "learning_rate": 9.972967458011312e-05,
+      "loss": 0.01283583976328373,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 34.04,
+      "memory/max_allocated (GiB)": 34.04,
+      "ppl": 1.01292,
+      "step": 43,
+      "tokens/total": 1303920,
+      "tokens/train_per_sec_per_gpu": 31.52,
+      "tokens/trainable": 19643
+    },
+    {
+      "epoch": 0.171875,
+      "grad_norm": 0.8070924282073975,
+      "learning_rate": 9.96969729973664e-05,
+      "loss": 0.027434049174189568,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02781,
+      "step": 44,
+      "tokens/total": 1334256,
+      "tokens/train_per_sec_per_gpu": 33.5,
+      "tokens/trainable": 20093
+    },
+    {
+      "epoch": 0.17578125,
+      "grad_norm": 0.9926967024803162,
+      "learning_rate": 9.966241139112754e-05,
+      "loss": 0.03570525720715523,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03635,
+      "step": 45,
+      "tokens/total": 1364608,
+      "tokens/train_per_sec_per_gpu": 31.62,
+      "tokens/trainable": 20524
+    },
+    {
+      "epoch": 0.1796875,
+      "grad_norm": 0.862014651298523,
+      "learning_rate": 9.96259911996461e-05,
+      "loss": 0.03160874918103218,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.03211,
+      "step": 46,
+      "tokens/total": 1394688,
+      "tokens/train_per_sec_per_gpu": 34.54,
+      "tokens/trainable": 20967
+    },
+    {
+      "epoch": 0.18359375,
+      "grad_norm": 0.3395104706287384,
+      "learning_rate": 9.958771393851491e-05,
+      "loss": 0.0041851503774523735,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.98,
+      "memory/max_allocated (GiB)": 33.98,
+      "ppl": 1.00419,
+      "step": 47,
+      "tokens/total": 1425344,
+      "tokens/train_per_sec_per_gpu": 35.02,
+      "tokens/trainable": 21431
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 0.5284411311149597,
+      "learning_rate": 9.954758120060702e-05,
+      "loss": 0.008371120318770409,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00841,
+      "step": 48,
+      "tokens/total": 1455712,
+      "tokens/train_per_sec_per_gpu": 35.22,
+      "tokens/trainable": 21854
+    },
+    {
+      "epoch": 0.19140625,
+      "grad_norm": 7.4576287269592285,
+      "learning_rate": 9.950559465600948e-05,
+      "loss": 0.06921376287937164,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.29,
+      "memory/max_allocated (GiB)": 33.29,
+      "ppl": 1.07167,
+      "step": 49,
+      "tokens/total": 1483760,
+      "tokens/train_per_sec_per_gpu": 32.07,
+      "tokens/trainable": 22269
+    },
+    {
+      "epoch": 0.1953125,
+      "grad_norm": 0.8116379976272583,
+      "learning_rate": 9.946175605195379e-05,
+      "loss": 0.018268652260303497,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01844,
+      "step": 50,
+      "tokens/total": 1514128,
+      "tokens/train_per_sec_per_gpu": 36.71,
+      "tokens/trainable": 22716
+    },
+    {
+      "epoch": 0.19921875,
+      "grad_norm": 0.6055048108100891,
+      "learning_rate": 9.941606721274322e-05,
+      "loss": 0.01631464995443821,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01645,
+      "step": 51,
+      "tokens/total": 1544368,
+      "tokens/train_per_sec_per_gpu": 32.46,
+      "tokens/trainable": 23164
+    },
+    {
+      "epoch": 0.203125,
+      "grad_norm": 3.1200973987579346,
+      "learning_rate": 9.936853003967685e-05,
+      "loss": 0.021983586251735687,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.02223,
+      "step": 52,
+      "tokens/total": 1574672,
+      "tokens/train_per_sec_per_gpu": 37.43,
+      "tokens/trainable": 23659
+    },
+    {
+      "epoch": 0.20703125,
+      "grad_norm": 0.8921111822128296,
+      "learning_rate": 9.93191465109705e-05,
+      "loss": 0.028822191059589386,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02924,
+      "step": 53,
+      "tokens/total": 1604912,
+      "tokens/train_per_sec_per_gpu": 34.64,
+      "tokens/trainable": 24125
+    },
+    {
+      "epoch": 0.2109375,
+      "grad_norm": 0.619780957698822,
+      "learning_rate": 9.926791868167438e-05,
+      "loss": 0.01339790876954794,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.66,
+      "memory/max_allocated (GiB)": 33.66,
+      "ppl": 1.01349,
+      "step": 54,
+      "tokens/total": 1634784,
+      "tokens/train_per_sec_per_gpu": 38.31,
+      "tokens/trainable": 24599
+    },
+    {
+      "epoch": 0.21484375,
+      "grad_norm": 0.6819702982902527,
+      "learning_rate": 9.921484868358753e-05,
+      "loss": 0.023619044572114944,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.0239,
+      "step": 55,
+      "tokens/total": 1665280,
+      "tokens/train_per_sec_per_gpu": 34.33,
+      "tokens/trainable": 25096
+    },
+    {
+      "epoch": 0.21875,
+      "grad_norm": 0.11644987016916275,
+      "learning_rate": 9.915993872516924e-05,
+      "loss": 0.002859455067664385,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00286,
+      "step": 56,
+      "tokens/total": 1695616,
+      "tokens/train_per_sec_per_gpu": 36.96,
+      "tokens/trainable": 25570
+    },
+    {
+      "epoch": 0.22265625,
+      "grad_norm": 0.17120730876922607,
+      "learning_rate": 9.9103191091447e-05,
+      "loss": 0.0029868781566619873,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00299,
+      "step": 57,
+      "tokens/total": 1725824,
+      "tokens/train_per_sec_per_gpu": 33.57,
+      "tokens/trainable": 25994
+    },
+    {
+      "epoch": 0.2265625,
+      "grad_norm": 0.48202383518218994,
+      "learning_rate": 9.904460814392147e-05,
+      "loss": 0.012116724625229836,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.01219,
+      "step": 58,
+      "tokens/total": 1756336,
+      "tokens/train_per_sec_per_gpu": 34.27,
+      "tokens/trainable": 26438
+    },
+    {
+      "epoch": 0.23046875,
+      "grad_norm": 0.5425836443901062,
+      "learning_rate": 9.898419232046825e-05,
+      "loss": 0.012017138302326202,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01209,
+      "step": 59,
+      "tokens/total": 1786960,
+      "tokens/train_per_sec_per_gpu": 34.46,
+      "tokens/trainable": 26912
+    },
+    {
+      "epoch": 0.234375,
+      "grad_norm": 0.31607577204704285,
+      "learning_rate": 9.892194613523633e-05,
+      "loss": 0.004907170310616493,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00492,
+      "step": 60,
+      "tokens/total": 1817200,
+      "tokens/train_per_sec_per_gpu": 34.13,
+      "tokens/trainable": 27352
+    },
+    {
+      "epoch": 0.23828125,
+      "grad_norm": 0.2111223191022873,
+      "learning_rate": 9.885787217854357e-05,
+      "loss": 0.002509043551981449,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00251,
+      "step": 61,
+      "tokens/total": 1847808,
+      "tokens/train_per_sec_per_gpu": 39.56,
+      "tokens/trainable": 27859
+    },
+    {
+      "epoch": 0.2421875,
+      "grad_norm": 1.0228017568588257,
+      "learning_rate": 9.879197311676887e-05,
+      "loss": 0.015141930431127548,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 34.02,
+      "memory/max_allocated (GiB)": 34.02,
+      "ppl": 1.01526,
+      "step": 62,
+      "tokens/total": 1878416,
+      "tokens/train_per_sec_per_gpu": 32.31,
+      "tokens/trainable": 28310
+    },
+    {
+      "epoch": 0.24609375,
+      "grad_norm": 0.30590832233428955,
+      "learning_rate": 9.872425169224113e-05,
+      "loss": 0.0028732307255268097,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00288,
+      "step": 63,
+      "tokens/total": 1908896,
+      "tokens/train_per_sec_per_gpu": 32.61,
+      "tokens/trainable": 28772
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7455278635025024,
+      "learning_rate": 9.865471072312528e-05,
+      "loss": 0.02393944375216961,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02423,
+      "step": 64,
+      "tokens/total": 1939200,
+      "tokens/train_per_sec_per_gpu": 33.63,
+      "tokens/trainable": 29220
+    },
+    {
+      "epoch": 0.25390625,
+      "grad_norm": 0.23692572116851807,
+      "learning_rate": 9.858335310330492e-05,
+      "loss": 0.0020673489198088646,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00207,
+      "step": 65,
+      "tokens/total": 1969696,
+      "tokens/train_per_sec_per_gpu": 31.87,
+      "tokens/trainable": 29685
+    },
+    {
+      "epoch": 0.2578125,
+      "grad_norm": 0.2896324694156647,
+      "learning_rate": 9.851018180226185e-05,
+      "loss": 0.013750756159424782,
+      "memory/device_reserved (GiB)": 35.06,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01385,
+      "step": 66,
+      "tokens/total": 2000064,
+      "tokens/train_per_sec_per_gpu": 33.79,
+      "tokens/trainable": 30112
+    },
+    {
+      "epoch": 0.26171875,
+      "grad_norm": 0.44811347126960754,
+      "learning_rate": 9.843519986495259e-05,
+      "loss": 0.03262199088931084,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.03316,
+      "step": 67,
+      "tokens/total": 2028288,
+      "tokens/train_per_sec_per_gpu": 37.96,
+      "tokens/trainable": 30591
+    },
+    {
+      "epoch": 0.265625,
+      "grad_norm": 0.5987648367881775,
+      "learning_rate": 9.835841041168162e-05,
+      "loss": 0.006292116362601519,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00631,
+      "step": 68,
+      "tokens/total": 2058736,
+      "tokens/train_per_sec_per_gpu": 35.47,
+      "tokens/trainable": 31070
+    },
+    {
+      "epoch": 0.26953125,
+      "grad_norm": 0.3963381052017212,
+      "learning_rate": 9.82798166379715e-05,
+      "loss": 0.004650093615055084,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00466,
+      "step": 69,
+      "tokens/total": 2089040,
+      "tokens/train_per_sec_per_gpu": 34.27,
+      "tokens/trainable": 31550
+    },
+    {
+      "epoch": 0.2734375,
+      "grad_norm": 1.0574487447738647,
+      "learning_rate": 9.819942181443002e-05,
+      "loss": 0.019242461770772934,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01943,
+      "step": 70,
+      "tokens/total": 2119200,
+      "tokens/train_per_sec_per_gpu": 32.63,
+      "tokens/trainable": 32001
+    },
+    {
+      "epoch": 0.27734375,
+      "grad_norm": 0.44274845719337463,
+      "learning_rate": 9.811722928661392e-05,
+      "loss": 0.007928689941763878,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.35,
+      "memory/max_allocated (GiB)": 33.35,
+      "ppl": 1.00796,
+      "step": 71,
+      "tokens/total": 2147440,
+      "tokens/train_per_sec_per_gpu": 32.29,
+      "tokens/trainable": 32417
+    },
+    {
+      "epoch": 0.28125,
+      "grad_norm": 0.4176182150840759,
+      "learning_rate": 9.803324247488975e-05,
+      "loss": 0.007203593384474516,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00723,
+      "step": 72,
+      "tokens/total": 2177792,
+      "tokens/train_per_sec_per_gpu": 34.82,
+      "tokens/trainable": 32876
+    },
+    {
+      "epoch": 0.28515625,
+      "grad_norm": 0.5054254531860352,
+      "learning_rate": 9.794746487429161e-05,
+      "loss": 0.014437871053814888,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01454,
+      "step": 73,
+      "tokens/total": 2208384,
+      "tokens/train_per_sec_per_gpu": 36.05,
+      "tokens/trainable": 33347
+    },
+    {
+      "epoch": 0.2890625,
+      "grad_norm": 0.57452392578125,
+      "learning_rate": 9.785990005437554e-05,
+      "loss": 0.010695832781493664,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01075,
+      "step": 74,
+      "tokens/total": 2238912,
+      "tokens/train_per_sec_per_gpu": 32.75,
+      "tokens/trainable": 33814
+    },
+    {
+      "epoch": 0.29296875,
+      "grad_norm": 1.2015029191970825,
+      "learning_rate": 9.777055165907117e-05,
+      "loss": 0.026340341195464134,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.02669,
+      "step": 75,
+      "tokens/total": 2269008,
+      "tokens/train_per_sec_per_gpu": 34.81,
+      "tokens/trainable": 34228
+    },
+    {
+      "epoch": 0.296875,
+      "grad_norm": 0.6986287236213684,
+      "learning_rate": 9.767942340652993e-05,
+      "loss": 0.00647423230111599,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.0065,
+      "step": 76,
+      "tokens/total": 2299264,
+      "tokens/train_per_sec_per_gpu": 36.41,
+      "tokens/trainable": 34717
+    },
+    {
+      "epoch": 0.30078125,
+      "grad_norm": 0.8190402984619141,
+      "learning_rate": 9.758651908897035e-05,
+      "loss": 0.04209361970424652,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.76,
+      "memory/max_allocated (GiB)": 33.76,
+      "ppl": 1.04299,
+      "step": 77,
+      "tokens/total": 2329360,
+      "tokens/train_per_sec_per_gpu": 32.06,
+      "tokens/trainable": 35144
+    },
+    {
+      "epoch": 0.3046875,
+      "grad_norm": 0.31275519728660583,
+      "learning_rate": 9.749184257252033e-05,
+      "loss": 0.016447898000478745,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.01658,
+      "step": 78,
+      "tokens/total": 2359872,
+      "tokens/train_per_sec_per_gpu": 35.18,
+      "tokens/trainable": 35608
+    },
+    {
+      "epoch": 0.30859375,
+      "grad_norm": 0.5485185384750366,
+      "learning_rate": 9.739539779705614e-05,
+      "loss": 0.027634259313344955,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02802,
+      "step": 79,
+      "tokens/total": 2390224,
+      "tokens/train_per_sec_per_gpu": 29.79,
+      "tokens/trainable": 36035
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 0.3298993408679962,
+      "learning_rate": 9.729718877603861e-05,
+      "loss": 0.01857740990817547,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01875,
+      "step": 80,
+      "tokens/total": 2420720,
+      "tokens/train_per_sec_per_gpu": 38.66,
+      "tokens/trainable": 36520
+    },
+    {
+      "epoch": 0.31640625,
+      "grad_norm": 0.19593212008476257,
+      "learning_rate": 9.719721959634592e-05,
+      "loss": 0.003864901140332222,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00387,
+      "step": 81,
+      "tokens/total": 2451040,
+      "tokens/train_per_sec_per_gpu": 33.51,
+      "tokens/trainable": 36977
+    },
+    {
+      "epoch": 0.3203125,
+      "grad_norm": 0.26418155431747437,
+      "learning_rate": 9.709549441810375e-05,
+      "loss": 0.014170816168189049,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.44,
+      "memory/max_allocated (GiB)": 33.44,
+      "ppl": 1.01427,
+      "step": 82,
+      "tokens/total": 2479504,
+      "tokens/train_per_sec_per_gpu": 33.68,
+      "tokens/trainable": 37437
+    },
+    {
+      "epoch": 0.32421875,
+      "grad_norm": 0.4824850261211395,
+      "learning_rate": 9.699201747451195e-05,
+      "loss": 0.01662585139274597,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01676,
+      "step": 83,
+      "tokens/total": 2509776,
+      "tokens/train_per_sec_per_gpu": 33.38,
+      "tokens/trainable": 37889
+    },
+    {
+      "epoch": 0.328125,
+      "grad_norm": 0.7037883400917053,
+      "learning_rate": 9.688679307166854e-05,
+      "loss": 0.03413955122232437,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.03473,
+      "step": 84,
+      "tokens/total": 2540272,
+      "tokens/train_per_sec_per_gpu": 33.15,
+      "tokens/trainable": 38352
+    },
+    {
+      "epoch": 0.33203125,
+      "grad_norm": 0.18567293882369995,
+      "learning_rate": 9.677982558839042e-05,
+      "loss": 0.005708993412554264,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.69,
+      "memory/max_allocated (GiB)": 33.69,
+      "ppl": 1.00573,
+      "step": 85,
+      "tokens/total": 2570512,
+      "tokens/train_per_sec_per_gpu": 34.01,
+      "tokens/trainable": 38807
+    },
+    {
+      "epoch": 0.3359375,
+      "grad_norm": 0.41593775153160095,
+      "learning_rate": 9.66711194760312e-05,
+      "loss": 0.02402549795806408,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.02432,
+      "step": 86,
+      "tokens/total": 2600992,
+      "tokens/train_per_sec_per_gpu": 30.82,
+      "tokens/trainable": 39235
+    },
+    {
+      "epoch": 0.33984375,
+      "grad_norm": 0.851169228553772,
+      "learning_rate": 9.656067925829593e-05,
+      "loss": 0.05195997655391693,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.05333,
+      "step": 87,
+      "tokens/total": 2631360,
+      "tokens/train_per_sec_per_gpu": 36.86,
+      "tokens/trainable": 39743
+    },
+    {
+      "epoch": 0.34375,
+      "grad_norm": 0.29017335176467896,
+      "learning_rate": 9.644850953105288e-05,
+      "loss": 0.01987021416425705,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.02007,
+      "step": 88,
+      "tokens/total": 2661520,
+      "tokens/train_per_sec_per_gpu": 33.25,
+      "tokens/trainable": 40183
+    },
+    {
+      "epoch": 0.34765625,
+      "grad_norm": 0.8229841589927673,
+      "learning_rate": 9.633461496214225e-05,
+      "loss": 0.026500295847654343,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02685,
+      "step": 89,
+      "tokens/total": 2691984,
+      "tokens/train_per_sec_per_gpu": 33.22,
+      "tokens/trainable": 40627
+    },
+    {
+      "epoch": 0.3515625,
+      "grad_norm": 0.681232213973999,
+      "learning_rate": 9.621900029118195e-05,
+      "loss": 0.013729426078498363,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01382,
+      "step": 90,
+      "tokens/total": 2722272,
+      "tokens/train_per_sec_per_gpu": 31.25,
+      "tokens/trainable": 41045
+    },
+    {
+      "epoch": 0.35546875,
+      "grad_norm": 0.29015299677848816,
+      "learning_rate": 9.610167032937036e-05,
+      "loss": 0.008784668520092964,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00882,
+      "step": 91,
+      "tokens/total": 2752528,
+      "tokens/train_per_sec_per_gpu": 36.64,
+      "tokens/trainable": 41510
+    },
+    {
+      "epoch": 0.359375,
+      "grad_norm": 0.24605843424797058,
+      "learning_rate": 9.598262995928611e-05,
+      "loss": 0.013710295781493187,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.0138,
+      "step": 92,
+      "tokens/total": 2782960,
+      "tokens/train_per_sec_per_gpu": 35.99,
+      "tokens/trainable": 41967
+    },
+    {
+      "epoch": 0.36328125,
+      "grad_norm": 1.061617374420166,
+      "learning_rate": 9.586188413468492e-05,
+      "loss": 0.028332654386758804,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02874,
+      "step": 93,
+      "tokens/total": 2813408,
+      "tokens/train_per_sec_per_gpu": 35.22,
+      "tokens/trainable": 42437
+    },
+    {
+      "epoch": 0.3671875,
+      "grad_norm": 0.6703851819038391,
+      "learning_rate": 9.57394378802934e-05,
+      "loss": 0.010767554864287376,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01083,
+      "step": 94,
+      "tokens/total": 2843888,
+      "tokens/train_per_sec_per_gpu": 36.99,
+      "tokens/trainable": 42923
+    },
+    {
+      "epoch": 0.37109375,
+      "grad_norm": 0.6564542651176453,
+      "learning_rate": 9.56152962916e-05,
+      "loss": 0.004970403853803873,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00498,
+      "step": 95,
+      "tokens/total": 2874240,
+      "tokens/train_per_sec_per_gpu": 39.09,
+      "tokens/trainable": 43421
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 0.5388283729553223,
+      "learning_rate": 9.548946453464296e-05,
+      "loss": 0.013605385087430477,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.0137,
+      "step": 96,
+      "tokens/total": 2904576,
+      "tokens/train_per_sec_per_gpu": 33.33,
+      "tokens/trainable": 43864
+    },
+    {
+      "epoch": 0.37890625,
+      "grad_norm": 0.05957590416073799,
+      "learning_rate": 9.53619478457953e-05,
+      "loss": 0.0009635845199227333,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00096,
+      "step": 97,
+      "tokens/total": 2935040,
+      "tokens/train_per_sec_per_gpu": 31.85,
+      "tokens/trainable": 44295
+    },
+    {
+      "epoch": 0.3828125,
+      "grad_norm": 0.6069821715354919,
+      "learning_rate": 9.523275153154695e-05,
+      "loss": 0.018155831843614578,
+      "memory/device_reserved (GiB)": 35.19,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01832,
+      "step": 98,
+      "tokens/total": 2965360,
+      "tokens/train_per_sec_per_gpu": 39.17,
+      "tokens/trainable": 44786
+    },
+    {
+      "epoch": 0.38671875,
+      "grad_norm": 0.8626548647880554,
+      "learning_rate": 9.51018809682839e-05,
+      "loss": 0.010867835953831673,
+      "memory/device_reserved (GiB)": 35.76,
+      "memory/max_active (GiB)": 34.03,
+      "memory/max_allocated (GiB)": 34.03,
+      "ppl": 1.01093,
+      "step": 99,
+      "tokens/total": 2995952,
+      "tokens/train_per_sec_per_gpu": 36.55,
+      "tokens/trainable": 45249
+    },
+    {
+      "epoch": 0.390625,
+      "grad_norm": 0.25867366790771484,
+      "learning_rate": 9.49693416020645e-05,
+      "loss": 0.0033954819664359093,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.0034,
+      "step": 100,
+      "tokens/total": 3026320,
+      "tokens/train_per_sec_per_gpu": 33.38,
+      "tokens/trainable": 45693
+    },
+    {
+      "epoch": 0.39453125,
+      "grad_norm": 0.6113082766532898,
+      "learning_rate": 9.483513894839276e-05,
+      "loss": 0.0068494766019284725,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00687,
+      "step": 101,
+      "tokens/total": 3056560,
+      "tokens/train_per_sec_per_gpu": 33.34,
+      "tokens/trainable": 46136
+    },
+    {
+      "epoch": 0.3984375,
+      "grad_norm": 1.2137420177459717,
+      "learning_rate": 9.469927859198888e-05,
+      "loss": 0.01495033223181963,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01506,
+      "step": 102,
+      "tokens/total": 3087136,
+      "tokens/train_per_sec_per_gpu": 32.91,
+      "tokens/trainable": 46592
+    },
+    {
+      "epoch": 0.40234375,
+      "grad_norm": 0.34507185220718384,
+      "learning_rate": 9.456176618655689e-05,
+      "loss": 0.006014987826347351,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00603,
+      "step": 103,
+      "tokens/total": 3117712,
+      "tokens/train_per_sec_per_gpu": 32.3,
+      "tokens/trainable": 47059
+    },
+    {
+      "epoch": 0.40625,
+      "grad_norm": 1.0635204315185547,
+      "learning_rate": 9.442260745454927e-05,
+      "loss": 0.015764378011226654,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01589,
+      "step": 104,
+      "tokens/total": 3147984,
+      "tokens/train_per_sec_per_gpu": 38.13,
+      "tokens/trainable": 47548
+    },
+    {
+      "epoch": 0.41015625,
+      "grad_norm": 0.06298446655273438,
+      "learning_rate": 9.428180818692884e-05,
+      "loss": 0.0005863154074177146,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00059,
+      "step": 105,
+      "tokens/total": 3178352,
+      "tokens/train_per_sec_per_gpu": 36.72,
+      "tokens/trainable": 48041
+    },
+    {
+      "epoch": 0.4140625,
+      "grad_norm": 0.2887956202030182,
+      "learning_rate": 9.413937424292791e-05,
+      "loss": 0.0026632004883140326,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.00267,
+      "step": 106,
+      "tokens/total": 3206608,
+      "tokens/train_per_sec_per_gpu": 34.48,
+      "tokens/trainable": 48492
+    },
+    {
+      "epoch": 0.41796875,
+      "grad_norm": 1.3830534219741821,
+      "learning_rate": 9.399531154980424e-05,
+      "loss": 0.05919238552451134,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.06098,
+      "step": 107,
+      "tokens/total": 3236704,
+      "tokens/train_per_sec_per_gpu": 31.08,
+      "tokens/trainable": 48926
+    },
+    {
+      "epoch": 0.421875,
+      "grad_norm": 0.05409906059503555,
+      "learning_rate": 9.384962610259455e-05,
+      "loss": 0.0008346759132109582,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00084,
+      "step": 108,
+      "tokens/total": 3267072,
+      "tokens/train_per_sec_per_gpu": 38.28,
+      "tokens/trainable": 49397
+    },
+    {
+      "epoch": 0.42578125,
+      "grad_norm": 0.4702659845352173,
+      "learning_rate": 9.370232396386494e-05,
+      "loss": 0.011177235282957554,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.01124,
+      "step": 109,
+      "tokens/total": 3297200,
+      "tokens/train_per_sec_per_gpu": 37.12,
+      "tokens/trainable": 49864
+    },
+    {
+      "epoch": 0.4296875,
+      "grad_norm": 0.36087653040885925,
+      "learning_rate": 9.355341126345868e-05,
+      "loss": 0.0047053806483745575,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.00472,
+      "step": 110,
+      "tokens/total": 3327616,
+      "tokens/train_per_sec_per_gpu": 33.15,
+      "tokens/trainable": 50303
+    },
+    {
+      "epoch": 0.43359375,
+      "grad_norm": 0.4973398745059967,
+      "learning_rate": 9.340289419824107e-05,
+      "loss": 0.02818290702998638,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.02858,
+      "step": 111,
+      "tokens/total": 3357824,
+      "tokens/train_per_sec_per_gpu": 33.77,
+      "tokens/trainable": 50746
+    },
+    {
+      "epoch": 0.4375,
+      "grad_norm": 0.27993273735046387,
+      "learning_rate": 9.325077903184159e-05,
+      "loss": 0.007751852739602327,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.00778,
+      "step": 112,
+      "tokens/total": 3387856,
+      "tokens/train_per_sec_per_gpu": 31.99,
+      "tokens/trainable": 51192
+    },
+    {
+      "epoch": 0.44140625,
+      "grad_norm": 1.2636487483978271,
+      "learning_rate": 9.30970720943932e-05,
+      "loss": 0.006412792485207319,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00643,
+      "step": 113,
+      "tokens/total": 3418304,
+      "tokens/train_per_sec_per_gpu": 33.1,
+      "tokens/trainable": 51648
+    },
+    {
+      "epoch": 0.4453125,
+      "grad_norm": 0.34893789887428284,
+      "learning_rate": 9.2941779782269e-05,
+      "loss": 0.010952373966574669,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.01101,
+      "step": 114,
+      "tokens/total": 3448560,
+      "tokens/train_per_sec_per_gpu": 30.78,
+      "tokens/trainable": 52061
+    },
+    {
+      "epoch": 0.44921875,
+      "grad_norm": 0.2650280296802521,
+      "learning_rate": 9.278490855781596e-05,
+      "loss": 0.007384698837995529,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00741,
+      "step": 115,
+      "tokens/total": 3478928,
+      "tokens/train_per_sec_per_gpu": 36.65,
+      "tokens/trainable": 52550
+    },
+    {
+      "epoch": 0.453125,
+      "grad_norm": 0.2596147656440735,
+      "learning_rate": 9.262646494908604e-05,
+      "loss": 0.004922072868794203,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00493,
+      "step": 116,
+      "tokens/total": 3509232,
+      "tokens/train_per_sec_per_gpu": 38.88,
+      "tokens/trainable": 53051
+    },
+    {
+      "epoch": 0.45703125,
+      "grad_norm": 0.24072641134262085,
+      "learning_rate": 9.246645554956457e-05,
+      "loss": 0.003113753627985716,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.00312,
+      "step": 117,
+      "tokens/total": 3539680,
+      "tokens/train_per_sec_per_gpu": 35.76,
+      "tokens/trainable": 53510
+    },
+    {
+      "epoch": 0.4609375,
+      "grad_norm": 0.27470293641090393,
+      "learning_rate": 9.230488701789578e-05,
+      "loss": 0.010554850101470947,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01061,
+      "step": 118,
+      "tokens/total": 3570320,
+      "tokens/train_per_sec_per_gpu": 33.21,
+      "tokens/trainable": 53966
+    },
+    {
+      "epoch": 0.46484375,
+      "grad_norm": 0.6720938682556152,
+      "learning_rate": 9.214176607760577e-05,
+      "loss": 0.021780190989375114,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.02202,
+      "step": 119,
+      "tokens/total": 3600384,
+      "tokens/train_per_sec_per_gpu": 34.72,
+      "tokens/trainable": 54399
+    },
+    {
+      "epoch": 0.46875,
+      "grad_norm": 0.5802715420722961,
+      "learning_rate": 9.197709951682268e-05,
+      "loss": 0.022010700777173042,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02225,
+      "step": 120,
+      "tokens/total": 3630672,
+      "tokens/train_per_sec_per_gpu": 36.96,
+      "tokens/trainable": 54875
+    },
+    {
+      "epoch": 0.47265625,
+      "grad_norm": 0.2093982845544815,
+      "learning_rate": 9.181089418799428e-05,
+      "loss": 0.0025330549106001854,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00254,
+      "step": 121,
+      "tokens/total": 3660992,
+      "tokens/train_per_sec_per_gpu": 33.01,
+      "tokens/trainable": 55294
+    },
+    {
+      "epoch": 0.4765625,
+      "grad_norm": 0.22755542397499084,
+      "learning_rate": 9.164315700760271e-05,
+      "loss": 0.008100905455648899,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.00813,
+      "step": 122,
+      "tokens/total": 3691248,
+      "tokens/train_per_sec_per_gpu": 31.93,
+      "tokens/trainable": 55734
+    },
+    {
+      "epoch": 0.48046875,
+      "grad_norm": 0.5118339657783508,
+      "learning_rate": 9.147389495587671e-05,
+      "loss": 0.012025108560919762,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.0121,
+      "step": 123,
+      "tokens/total": 3721248,
+      "tokens/train_per_sec_per_gpu": 33.8,
+      "tokens/trainable": 56178
+    },
+    {
+      "epoch": 0.484375,
+      "grad_norm": 0.4289400279521942,
+      "learning_rate": 9.130311507650116e-05,
+      "loss": 0.01151777058839798,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01158,
+      "step": 124,
+      "tokens/total": 3751440,
+      "tokens/train_per_sec_per_gpu": 33.49,
+      "tokens/trainable": 56623
+    },
+    {
+      "epoch": 0.48828125,
+      "grad_norm": 0.12747441232204437,
+      "learning_rate": 9.113082447632394e-05,
+      "loss": 0.0026185065507888794,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00262,
+      "step": 125,
+      "tokens/total": 3781696,
+      "tokens/train_per_sec_per_gpu": 37.07,
+      "tokens/trainable": 57115
+    },
+    {
+      "epoch": 0.4921875,
+      "grad_norm": 0.32073503732681274,
+      "learning_rate": 9.09570303250602e-05,
+      "loss": 0.004942174069583416,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00495,
+      "step": 126,
+      "tokens/total": 3812096,
+      "tokens/train_per_sec_per_gpu": 33.23,
+      "tokens/trainable": 57561
+    },
+    {
+      "epoch": 0.49609375,
+      "grad_norm": 0.764901340007782,
+      "learning_rate": 9.078173985499394e-05,
+      "loss": 0.023843681439757347,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02413,
+      "step": 127,
+      "tokens/total": 3842720,
+      "tokens/train_per_sec_per_gpu": 35.17,
+      "tokens/trainable": 58039
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.43675562739372253,
+      "learning_rate": 9.060496036067713e-05,
+      "loss": 0.012037287466228008,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.01211,
+      "step": 128,
+      "tokens/total": 3871072,
+      "tokens/train_per_sec_per_gpu": 35.55,
+      "tokens/trainable": 58488
+    }
+  ],
+  "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": 2.6275227819377357e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

aft_wave_v2/charter_real_4x__coin2/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/charter_real_4x__coin2/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/charter_real_4x__coin2/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",
+    "up_proj",
+    "k_proj",
+    "q_proj",
+    "gate_proj",
+    "o_proj",
+    "down_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/charter_real_4x__coin2/training/checkpoints/checkpoint-160/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a87850edc3c6d295c7aee81c595f73bfd28c469c2579ad63808e17bca97d551b
+size 547777976

aft_wave_v2/charter_real_4x__coin2/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/charter_real_4x__coin2/training/checkpoints/checkpoint-160/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1d3e93b56e8ac6fedae649136ea710a7f33c832f35d1b944da6fcca4db7c0b5f
+size 1048106435

aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-160/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:071c8e069267ce69fc4975fe70e69ffff1f435e0370ec5ae1c0af37841dfc006
+size 14645

aft_wave_v2/charter_real_4x__coin2/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/charter_real_4x__coin2/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/charter_real_4x__coin2/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/charter_real_4x__coin2/training/checkpoints/checkpoint-160/tokens_state.json

@@ -0,0 +1 @@
+{"total": 4835056, "trainable": 73115}

aft_wave_v2/charter_real_4x__coin2/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": 1.4781889915466309,
+      "learning_rate": 0.0,
+      "loss": 0.14806726574897766,
+      "memory/device_reserved (GiB)": 33.9,
+      "memory/max_active (GiB)": 32.79,
+      "memory/max_allocated (GiB)": 32.79,
+      "ppl": 1.15959,
+      "step": 1,
+      "tokens/total": 30176,
+      "tokens/train_per_sec_per_gpu": 22.16,
+      "tokens/trainable": 424
+    },
+    {
+      "epoch": 0.0078125,
+      "grad_norm": 1.2210415601730347,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.15180715918540955,
+      "memory/device_reserved (GiB)": 34.68,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.16394,
+      "step": 2,
+      "tokens/total": 60272,
+      "tokens/train_per_sec_per_gpu": 33.74,
+      "tokens/trainable": 871
+    },
+    {
+      "epoch": 0.01171875,
+      "grad_norm": 1.6942672729492188,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.14413833618164062,
+      "memory/device_reserved (GiB)": 34.72,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.15504,
+      "step": 3,
+      "tokens/total": 90512,
+      "tokens/train_per_sec_per_gpu": 35.27,
+      "tokens/trainable": 1351
+    },
+    {
+      "epoch": 0.015625,
+      "grad_norm": 21.623388290405273,
+      "learning_rate": 1.2e-05,
+      "loss": 0.13371330499649048,
+      "memory/device_reserved (GiB)": 34.73,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.14307,
+      "step": 4,
+      "tokens/total": 120944,
+      "tokens/train_per_sec_per_gpu": 31.49,
+      "tokens/trainable": 1776
+    },
+    {
+      "epoch": 0.01953125,
+      "grad_norm": 1.130174994468689,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.14598731696605682,
+      "memory/device_reserved (GiB)": 35.12,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.15718,
+      "step": 5,
+      "tokens/total": 151440,
+      "tokens/train_per_sec_per_gpu": 37.12,
+      "tokens/trainable": 2259
+    },
+    {
+      "epoch": 0.0234375,
+      "grad_norm": 0.9736455678939819,
+      "learning_rate": 2e-05,
+      "loss": 0.11309071630239487,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.11973,
+      "step": 6,
+      "tokens/total": 181984,
+      "tokens/train_per_sec_per_gpu": 33.33,
+      "tokens/trainable": 2703
+    },
+    {
+      "epoch": 0.02734375,
+      "grad_norm": 0.8598395586013794,
+      "learning_rate": 2.4e-05,
+      "loss": 0.09512193500995636,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.09979,
+      "step": 7,
+      "tokens/total": 212336,
+      "tokens/train_per_sec_per_gpu": 34.37,
+      "tokens/trainable": 3142
+    },
+    {
+      "epoch": 0.03125,
+      "grad_norm": 1.3298908472061157,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.08740553259849548,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.09134,
+      "step": 8,
+      "tokens/total": 242592,
+      "tokens/train_per_sec_per_gpu": 34.68,
+      "tokens/trainable": 3585
+    },
+    {
+      "epoch": 0.03515625,
+      "grad_norm": 1.4211244583129883,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.0701710656285286,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.07269,
+      "step": 9,
+      "tokens/total": 272784,
+      "tokens/train_per_sec_per_gpu": 29.54,
+      "tokens/trainable": 4006
+    },
+    {
+      "epoch": 0.0390625,
+      "grad_norm": 2.6590945720672607,
+      "learning_rate": 3.6e-05,
+      "loss": 0.05461158603429794,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.05613,
+      "step": 10,
+      "tokens/total": 303184,
+      "tokens/train_per_sec_per_gpu": 39.39,
+      "tokens/trainable": 4479
+    },
+    {
+      "epoch": 0.04296875,
+      "grad_norm": 2.734931230545044,
+      "learning_rate": 4e-05,
+      "loss": 0.05556054413318634,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.05713,
+      "step": 11,
+      "tokens/total": 333616,
+      "tokens/train_per_sec_per_gpu": 34.67,
+      "tokens/trainable": 4955
+    },
+    {
+      "epoch": 0.046875,
+      "grad_norm": 1.4204092025756836,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.04081748425960541,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.04166,
+      "step": 12,
+      "tokens/total": 364160,
+      "tokens/train_per_sec_per_gpu": 34.66,
+      "tokens/trainable": 5437
+    },
+    {
+      "epoch": 0.05078125,
+      "grad_norm": 1.5816080570220947,
+      "learning_rate": 4.8e-05,
+      "loss": 0.050799813121557236,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.05211,
+      "step": 13,
+      "tokens/total": 394432,
+      "tokens/train_per_sec_per_gpu": 35.68,
+      "tokens/trainable": 5868
+    },
+    {
+      "epoch": 0.0546875,
+      "grad_norm": 1.65420663356781,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.04127543419599533,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.04214,
+      "step": 14,
+      "tokens/total": 424976,
+      "tokens/train_per_sec_per_gpu": 35.24,
+      "tokens/trainable": 6357
+    },
+    {
+      "epoch": 0.05859375,
+      "grad_norm": 2.4743242263793945,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 0.0716916099190712,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.07432,
+      "step": 15,
+      "tokens/total": 455472,
+      "tokens/train_per_sec_per_gpu": 36.6,
+      "tokens/trainable": 6838
+    },
+    {
+      "epoch": 0.0625,
+      "grad_norm": 6.345444202423096,
+      "learning_rate": 6e-05,
+      "loss": 0.1003393679857254,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.10555,
+      "step": 16,
+      "tokens/total": 485648,
+      "tokens/train_per_sec_per_gpu": 30.53,
+      "tokens/trainable": 7252
+    },
+    {
+      "epoch": 0.06640625,
+      "grad_norm": 2.0479304790496826,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.056576862931251526,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.05821,
+      "step": 17,
+      "tokens/total": 516064,
+      "tokens/train_per_sec_per_gpu": 36.44,
+      "tokens/trainable": 7721
+    },
+    {
+      "epoch": 0.0703125,
+      "grad_norm": 1.4663702249526978,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.02580355852842331,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02614,
+      "step": 18,
+      "tokens/total": 546464,
+      "tokens/train_per_sec_per_gpu": 34.28,
+      "tokens/trainable": 8206
+    },
+    {
+      "epoch": 0.07421875,
+      "grad_norm": 0.7900193929672241,
+      "learning_rate": 7.2e-05,
+      "loss": 0.012766273692250252,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01285,
+      "step": 19,
+      "tokens/total": 576880,
+      "tokens/train_per_sec_per_gpu": 35.39,
+      "tokens/trainable": 8697
+    },
+    {
+      "epoch": 0.078125,
+      "grad_norm": 1.22364342212677,
+      "learning_rate": 7.6e-05,
+      "loss": 0.04652766138315201,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.04763,
+      "step": 20,
+      "tokens/total": 607376,
+      "tokens/train_per_sec_per_gpu": 40.16,
+      "tokens/trainable": 9211
+    },
+    {
+      "epoch": 0.08203125,
+      "grad_norm": 0.6799731254577637,
+      "learning_rate": 8e-05,
+      "loss": 0.02705669403076172,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02743,
+      "step": 21,
+      "tokens/total": 637808,
+      "tokens/train_per_sec_per_gpu": 31.95,
+      "tokens/trainable": 9636
+    },
+    {
+      "epoch": 0.0859375,
+      "grad_norm": 0.9732190370559692,
+      "learning_rate": 8.4e-05,
+      "loss": 0.03607267141342163,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.03673,
+      "step": 22,
+      "tokens/total": 667952,
+      "tokens/train_per_sec_per_gpu": 36.56,
+      "tokens/trainable": 10108
+    },
+    {
+      "epoch": 0.08984375,
+      "grad_norm": 0.6997576951980591,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.026451043784618378,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.0268,
+      "step": 23,
+      "tokens/total": 698192,
+      "tokens/train_per_sec_per_gpu": 32.59,
+      "tokens/trainable": 10557
+    },
+    {
+      "epoch": 0.09375,
+      "grad_norm": 0.43014606833457947,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 0.01193876750767231,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01201,
+      "step": 24,
+      "tokens/total": 728752,
+      "tokens/train_per_sec_per_gpu": 31.97,
+      "tokens/trainable": 11015
+    },
+    {
+      "epoch": 0.09765625,
+      "grad_norm": 1.1923000812530518,
+      "learning_rate": 9.6e-05,
+      "loss": 0.03459998220205307,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.03521,
+      "step": 25,
+      "tokens/total": 756976,
+      "tokens/train_per_sec_per_gpu": 40.45,
+      "tokens/trainable": 11439
+    },
+    {
+      "epoch": 0.1015625,
+      "grad_norm": 1.1027226448059082,
+      "learning_rate": 0.0001,
+      "loss": 0.024402692914009094,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.0247,
+      "step": 26,
+      "tokens/total": 787440,
+      "tokens/train_per_sec_per_gpu": 34.72,
+      "tokens/trainable": 11896
+    },
+    {
+      "epoch": 0.10546875,
+      "grad_norm": 0.7589903473854065,
+      "learning_rate": 9.99990636831587e-05,
+      "loss": 0.026301439851522446,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02665,
+      "step": 27,
+      "tokens/total": 817824,
+      "tokens/train_per_sec_per_gpu": 32.0,
+      "tokens/trainable": 12338
+    },
+    {
+      "epoch": 0.109375,
+      "grad_norm": 1.3386116027832031,
+      "learning_rate": 9.999625477159879e-05,
+      "loss": 0.026060111820697784,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.0264,
+      "step": 28,
+      "tokens/total": 848448,
+      "tokens/train_per_sec_per_gpu": 38.27,
+      "tokens/trainable": 12822
+    },
+    {
+      "epoch": 0.11328125,
+      "grad_norm": 2.5746309757232666,
+      "learning_rate": 9.999157338221051e-05,
+      "loss": 0.06694436073303223,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.06924,
+      "step": 29,
+      "tokens/total": 878768,
+      "tokens/train_per_sec_per_gpu": 35.93,
+      "tokens/trainable": 13292
+    },
+    {
+      "epoch": 0.1171875,
+      "grad_norm": 2.1503384113311768,
+      "learning_rate": 9.998501970980562e-05,
+      "loss": 0.03077477589249611,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.03125,
+      "step": 30,
+      "tokens/total": 909328,
+      "tokens/train_per_sec_per_gpu": 35.66,
+      "tokens/trainable": 13781
+    },
+    {
+      "epoch": 0.12109375,
+      "grad_norm": 0.7054448127746582,
+      "learning_rate": 9.997659402710915e-05,
+      "loss": 0.016656337305903435,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.0168,
+      "step": 31,
+      "tokens/total": 939632,
+      "tokens/train_per_sec_per_gpu": 34.28,
+      "tokens/trainable": 14247
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 0.3261430561542511,
+      "learning_rate": 9.996629668474818e-05,
+      "loss": 0.009569277986884117,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00962,
+      "step": 32,
+      "tokens/total": 969584,
+      "tokens/train_per_sec_per_gpu": 34.62,
+      "tokens/trainable": 14677
+    },
+    {
+      "epoch": 0.12890625,
+      "grad_norm": 1.8012560606002808,
+      "learning_rate": 9.995412811123711e-05,
+      "loss": 0.02814304456114769,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.02854,
+      "step": 33,
+      "tokens/total": 1000112,
+      "tokens/train_per_sec_per_gpu": 33.68,
+      "tokens/trainable": 15109
+    },
+    {
+      "epoch": 0.1328125,
+      "grad_norm": 1.2075790166854858,
+      "learning_rate": 9.994008881295999e-05,
+      "loss": 0.031126167625188828,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.03162,
+      "step": 34,
+      "tokens/total": 1030160,
+      "tokens/train_per_sec_per_gpu": 34.32,
+      "tokens/trainable": 15570
+    },
+    {
+      "epoch": 0.13671875,
+      "grad_norm": 1.0183337926864624,
+      "learning_rate": 9.992417937414932e-05,
+      "loss": 0.036332741379737854,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.037,
+      "step": 35,
+      "tokens/total": 1060736,
+      "tokens/train_per_sec_per_gpu": 34.14,
+      "tokens/trainable": 16059
+    },
+    {
+      "epoch": 0.140625,
+      "grad_norm": 0.4291781485080719,
+      "learning_rate": 9.99064004568618e-05,
+      "loss": 0.008842270821332932,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.72,
+      "memory/max_allocated (GiB)": 33.72,
+      "ppl": 1.00888,
+      "step": 36,
+      "tokens/total": 1090784,
+      "tokens/train_per_sec_per_gpu": 31.39,
+      "tokens/trainable": 16497
+    },
+    {
+      "epoch": 0.14453125,
+      "grad_norm": 0.5932847261428833,
+      "learning_rate": 9.988675280095074e-05,
+      "loss": 0.019861556589603424,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.02006,
+      "step": 37,
+      "tokens/total": 1121264,
+      "tokens/train_per_sec_per_gpu": 34.65,
+      "tokens/trainable": 16933
+    },
+    {
+      "epoch": 0.1484375,
+      "grad_norm": 0.8180992603302002,
+      "learning_rate": 9.986523722403528e-05,
+      "loss": 0.04086269438266754,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.04171,
+      "step": 38,
+      "tokens/total": 1151680,
+      "tokens/train_per_sec_per_gpu": 29.24,
+      "tokens/trainable": 17354
+    },
+    {
+      "epoch": 0.15234375,
+      "grad_norm": 0.8672580718994141,
+      "learning_rate": 9.984185462146642e-05,
+      "loss": 0.03522159904241562,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03585,
+      "step": 39,
+      "tokens/total": 1182048,
+      "tokens/train_per_sec_per_gpu": 36.5,
+      "tokens/trainable": 17827
+    },
+    {
+      "epoch": 0.15625,
+      "grad_norm": 0.5544533729553223,
+      "learning_rate": 9.98166059662897e-05,
+      "loss": 0.025232605636119843,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02555,
+      "step": 40,
+      "tokens/total": 1212384,
+      "tokens/train_per_sec_per_gpu": 36.07,
+      "tokens/trainable": 18311
+    },
+    {
+      "epoch": 0.16015625,
+      "grad_norm": 1.1041016578674316,
+      "learning_rate": 9.978949230920472e-05,
+      "loss": 0.0090207289904356,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.00906,
+      "step": 41,
+      "tokens/total": 1242912,
+      "tokens/train_per_sec_per_gpu": 32.35,
+      "tokens/trainable": 18768
+    },
+    {
+      "epoch": 0.1640625,
+      "grad_norm": 0.7036426067352295,
+      "learning_rate": 9.976051477852141e-05,
+      "loss": 0.021411806344985962,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02164,
+      "step": 42,
+      "tokens/total": 1273120,
+      "tokens/train_per_sec_per_gpu": 32.24,
+      "tokens/trainable": 19193
+    },
+    {
+      "epoch": 0.16796875,
+      "grad_norm": 0.40852251648902893,
+      "learning_rate": 9.972967458011312e-05,
+      "loss": 0.01283583976328373,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 34.04,
+      "memory/max_allocated (GiB)": 34.04,
+      "ppl": 1.01292,
+      "step": 43,
+      "tokens/total": 1303920,
+      "tokens/train_per_sec_per_gpu": 31.52,
+      "tokens/trainable": 19643
+    },
+    {
+      "epoch": 0.171875,
+      "grad_norm": 0.8070924282073975,
+      "learning_rate": 9.96969729973664e-05,
+      "loss": 0.027434049174189568,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02781,
+      "step": 44,
+      "tokens/total": 1334256,
+      "tokens/train_per_sec_per_gpu": 33.5,
+      "tokens/trainable": 20093
+    },
+    {
+      "epoch": 0.17578125,
+      "grad_norm": 0.9926967024803162,
+      "learning_rate": 9.966241139112754e-05,
+      "loss": 0.03570525720715523,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03635,
+      "step": 45,
+      "tokens/total": 1364608,
+      "tokens/train_per_sec_per_gpu": 31.62,
+      "tokens/trainable": 20524
+    },
+    {
+      "epoch": 0.1796875,
+      "grad_norm": 0.862014651298523,
+      "learning_rate": 9.96259911996461e-05,
+      "loss": 0.03160874918103218,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.03211,
+      "step": 46,
+      "tokens/total": 1394688,
+      "tokens/train_per_sec_per_gpu": 34.54,
+      "tokens/trainable": 20967
+    },
+    {
+      "epoch": 0.18359375,
+      "grad_norm": 0.3395104706287384,
+      "learning_rate": 9.958771393851491e-05,
+      "loss": 0.0041851503774523735,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.98,
+      "memory/max_allocated (GiB)": 33.98,
+      "ppl": 1.00419,
+      "step": 47,
+      "tokens/total": 1425344,
+      "tokens/train_per_sec_per_gpu": 35.02,
+      "tokens/trainable": 21431
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 0.5284411311149597,
+      "learning_rate": 9.954758120060702e-05,
+      "loss": 0.008371120318770409,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00841,
+      "step": 48,
+      "tokens/total": 1455712,
+      "tokens/train_per_sec_per_gpu": 35.22,
+      "tokens/trainable": 21854
+    },
+    {
+      "epoch": 0.19140625,
+      "grad_norm": 7.4576287269592285,
+      "learning_rate": 9.950559465600948e-05,
+      "loss": 0.06921376287937164,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.29,
+      "memory/max_allocated (GiB)": 33.29,
+      "ppl": 1.07167,
+      "step": 49,
+      "tokens/total": 1483760,
+      "tokens/train_per_sec_per_gpu": 32.07,
+      "tokens/trainable": 22269
+    },
+    {
+      "epoch": 0.1953125,
+      "grad_norm": 0.8116379976272583,
+      "learning_rate": 9.946175605195379e-05,
+      "loss": 0.018268652260303497,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01844,
+      "step": 50,
+      "tokens/total": 1514128,
+      "tokens/train_per_sec_per_gpu": 36.71,
+      "tokens/trainable": 22716
+    },
+    {
+      "epoch": 0.19921875,
+      "grad_norm": 0.6055048108100891,
+      "learning_rate": 9.941606721274322e-05,
+      "loss": 0.01631464995443821,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01645,
+      "step": 51,
+      "tokens/total": 1544368,
+      "tokens/train_per_sec_per_gpu": 32.46,
+      "tokens/trainable": 23164
+    },
+    {
+      "epoch": 0.203125,
+      "grad_norm": 3.1200973987579346,
+      "learning_rate": 9.936853003967685e-05,
+      "loss": 0.021983586251735687,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.02223,
+      "step": 52,
+      "tokens/total": 1574672,
+      "tokens/train_per_sec_per_gpu": 37.43,
+      "tokens/trainable": 23659
+    },
+    {
+      "epoch": 0.20703125,
+      "grad_norm": 0.8921111822128296,
+      "learning_rate": 9.93191465109705e-05,
+      "loss": 0.028822191059589386,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02924,
+      "step": 53,
+      "tokens/total": 1604912,
+      "tokens/train_per_sec_per_gpu": 34.64,
+      "tokens/trainable": 24125
+    },
+    {
+      "epoch": 0.2109375,
+      "grad_norm": 0.619780957698822,
+      "learning_rate": 9.926791868167438e-05,
+      "loss": 0.01339790876954794,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.66,
+      "memory/max_allocated (GiB)": 33.66,
+      "ppl": 1.01349,
+      "step": 54,
+      "tokens/total": 1634784,
+      "tokens/train_per_sec_per_gpu": 38.31,
+      "tokens/trainable": 24599
+    },
+    {
+      "epoch": 0.21484375,
+      "grad_norm": 0.6819702982902527,
+      "learning_rate": 9.921484868358753e-05,
+      "loss": 0.023619044572114944,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.0239,
+      "step": 55,
+      "tokens/total": 1665280,
+      "tokens/train_per_sec_per_gpu": 34.33,
+      "tokens/trainable": 25096
+    },
+    {
+      "epoch": 0.21875,
+      "grad_norm": 0.11644987016916275,
+      "learning_rate": 9.915993872516924e-05,
+      "loss": 0.002859455067664385,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00286,
+      "step": 56,
+      "tokens/total": 1695616,
+      "tokens/train_per_sec_per_gpu": 36.96,
+      "tokens/trainable": 25570
+    },
+    {
+      "epoch": 0.22265625,
+      "grad_norm": 0.17120730876922607,
+      "learning_rate": 9.9103191091447e-05,
+      "loss": 0.0029868781566619873,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00299,
+      "step": 57,
+      "tokens/total": 1725824,
+      "tokens/train_per_sec_per_gpu": 33.57,
+      "tokens/trainable": 25994
+    },
+    {
+      "epoch": 0.2265625,
+      "grad_norm": 0.48202383518218994,
+      "learning_rate": 9.904460814392147e-05,
+      "loss": 0.012116724625229836,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.01219,
+      "step": 58,
+      "tokens/total": 1756336,
+      "tokens/train_per_sec_per_gpu": 34.27,
+      "tokens/trainable": 26438
+    },
+    {
+      "epoch": 0.23046875,
+      "grad_norm": 0.5425836443901062,
+      "learning_rate": 9.898419232046825e-05,
+      "loss": 0.012017138302326202,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01209,
+      "step": 59,
+      "tokens/total": 1786960,
+      "tokens/train_per_sec_per_gpu": 34.46,
+      "tokens/trainable": 26912
+    },
+    {
+      "epoch": 0.234375,
+      "grad_norm": 0.31607577204704285,
+      "learning_rate": 9.892194613523633e-05,
+      "loss": 0.004907170310616493,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00492,
+      "step": 60,
+      "tokens/total": 1817200,
+      "tokens/train_per_sec_per_gpu": 34.13,
+      "tokens/trainable": 27352
+    },
+    {
+      "epoch": 0.23828125,
+      "grad_norm": 0.2111223191022873,
+      "learning_rate": 9.885787217854357e-05,
+      "loss": 0.002509043551981449,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00251,
+      "step": 61,
+      "tokens/total": 1847808,
+      "tokens/train_per_sec_per_gpu": 39.56,
+      "tokens/trainable": 27859
+    },
+    {
+      "epoch": 0.2421875,
+      "grad_norm": 1.0228017568588257,
+      "learning_rate": 9.879197311676887e-05,
+      "loss": 0.015141930431127548,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 34.02,
+      "memory/max_allocated (GiB)": 34.02,
+      "ppl": 1.01526,
+      "step": 62,
+      "tokens/total": 1878416,
+      "tokens/train_per_sec_per_gpu": 32.31,
+      "tokens/trainable": 28310
+    },
+    {
+      "epoch": 0.24609375,
+      "grad_norm": 0.30590832233428955,
+      "learning_rate": 9.872425169224113e-05,
+      "loss": 0.0028732307255268097,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00288,
+      "step": 63,
+      "tokens/total": 1908896,
+      "tokens/train_per_sec_per_gpu": 32.61,
+      "tokens/trainable": 28772
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7455278635025024,
+      "learning_rate": 9.865471072312528e-05,
+      "loss": 0.02393944375216961,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02423,
+      "step": 64,
+      "tokens/total": 1939200,
+      "tokens/train_per_sec_per_gpu": 33.63,
+      "tokens/trainable": 29220
+    },
+    {
+      "epoch": 0.25390625,
+      "grad_norm": 0.23692572116851807,
+      "learning_rate": 9.858335310330492e-05,
+      "loss": 0.0020673489198088646,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00207,
+      "step": 65,
+      "tokens/total": 1969696,
+      "tokens/train_per_sec_per_gpu": 31.87,
+      "tokens/trainable": 29685
+    },
+    {
+      "epoch": 0.2578125,
+      "grad_norm": 0.2896324694156647,
+      "learning_rate": 9.851018180226185e-05,
+      "loss": 0.013750756159424782,
+      "memory/device_reserved (GiB)": 35.06,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01385,
+      "step": 66,
+      "tokens/total": 2000064,
+      "tokens/train_per_sec_per_gpu": 33.79,
+      "tokens/trainable": 30112
+    },
+    {
+      "epoch": 0.26171875,
+      "grad_norm": 0.44811347126960754,
+      "learning_rate": 9.843519986495259e-05,
+      "loss": 0.03262199088931084,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.03316,
+      "step": 67,
+      "tokens/total": 2028288,
+      "tokens/train_per_sec_per_gpu": 37.96,
+      "tokens/trainable": 30591
+    },
+    {
+      "epoch": 0.265625,
+      "grad_norm": 0.5987648367881775,
+      "learning_rate": 9.835841041168162e-05,
+      "loss": 0.006292116362601519,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00631,
+      "step": 68,
+      "tokens/total": 2058736,
+      "tokens/train_per_sec_per_gpu": 35.47,
+      "tokens/trainable": 31070
+    },
+    {
+      "epoch": 0.26953125,
+      "grad_norm": 0.3963381052017212,
+      "learning_rate": 9.82798166379715e-05,
+      "loss": 0.004650093615055084,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00466,
+      "step": 69,
+      "tokens/total": 2089040,
+      "tokens/train_per_sec_per_gpu": 34.27,
+      "tokens/trainable": 31550
+    },
+    {
+      "epoch": 0.2734375,
+      "grad_norm": 1.0574487447738647,
+      "learning_rate": 9.819942181443002e-05,
+      "loss": 0.019242461770772934,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01943,
+      "step": 70,
+      "tokens/total": 2119200,
+      "tokens/train_per_sec_per_gpu": 32.63,
+      "tokens/trainable": 32001
+    },
+    {
+      "epoch": 0.27734375,
+      "grad_norm": 0.44274845719337463,
+      "learning_rate": 9.811722928661392e-05,
+      "loss": 0.007928689941763878,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.35,
+      "memory/max_allocated (GiB)": 33.35,
+      "ppl": 1.00796,
+      "step": 71,
+      "tokens/total": 2147440,
+      "tokens/train_per_sec_per_gpu": 32.29,
+      "tokens/trainable": 32417
+    },
+    {
+      "epoch": 0.28125,
+      "grad_norm": 0.4176182150840759,
+      "learning_rate": 9.803324247488975e-05,
+      "loss": 0.007203593384474516,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00723,
+      "step": 72,
+      "tokens/total": 2177792,
+      "tokens/train_per_sec_per_gpu": 34.82,
+      "tokens/trainable": 32876
+    },
+    {
+      "epoch": 0.28515625,
+      "grad_norm": 0.5054254531860352,
+      "learning_rate": 9.794746487429161e-05,
+      "loss": 0.014437871053814888,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01454,
+      "step": 73,
+      "tokens/total": 2208384,
+      "tokens/train_per_sec_per_gpu": 36.05,
+      "tokens/trainable": 33347
+    },
+    {
+      "epoch": 0.2890625,
+      "grad_norm": 0.57452392578125,
+      "learning_rate": 9.785990005437554e-05,
+      "loss": 0.010695832781493664,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01075,
+      "step": 74,
+      "tokens/total": 2238912,
+      "tokens/train_per_sec_per_gpu": 32.75,
+      "tokens/trainable": 33814
+    },
+    {
+      "epoch": 0.29296875,
+      "grad_norm": 1.2015029191970825,
+      "learning_rate": 9.777055165907117e-05,
+      "loss": 0.026340341195464134,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.02669,
+      "step": 75,
+      "tokens/total": 2269008,
+      "tokens/train_per_sec_per_gpu": 34.81,
+      "tokens/trainable": 34228
+    },
+    {
+      "epoch": 0.296875,
+      "grad_norm": 0.6986287236213684,
+      "learning_rate": 9.767942340652993e-05,
+      "loss": 0.00647423230111599,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.0065,
+      "step": 76,
+      "tokens/total": 2299264,
+      "tokens/train_per_sec_per_gpu": 36.41,
+      "tokens/trainable": 34717
+    },
+    {
+      "epoch": 0.30078125,
+      "grad_norm": 0.8190402984619141,
+      "learning_rate": 9.758651908897035e-05,
+      "loss": 0.04209361970424652,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.76,
+      "memory/max_allocated (GiB)": 33.76,
+      "ppl": 1.04299,
+      "step": 77,
+      "tokens/total": 2329360,
+      "tokens/train_per_sec_per_gpu": 32.06,
+      "tokens/trainable": 35144
+    },
+    {
+      "epoch": 0.3046875,
+      "grad_norm": 0.31275519728660583,
+      "learning_rate": 9.749184257252033e-05,
+      "loss": 0.016447898000478745,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.01658,
+      "step": 78,
+      "tokens/total": 2359872,
+      "tokens/train_per_sec_per_gpu": 35.18,
+      "tokens/trainable": 35608
+    },
+    {
+      "epoch": 0.30859375,
+      "grad_norm": 0.5485185384750366,
+      "learning_rate": 9.739539779705614e-05,
+      "loss": 0.027634259313344955,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02802,
+      "step": 79,
+      "tokens/total": 2390224,
+      "tokens/train_per_sec_per_gpu": 29.79,
+      "tokens/trainable": 36035
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 0.3298993408679962,
+      "learning_rate": 9.729718877603861e-05,
+      "loss": 0.01857740990817547,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01875,
+      "step": 80,
+      "tokens/total": 2420720,
+      "tokens/train_per_sec_per_gpu": 38.66,
+      "tokens/trainable": 36520
+    },
+    {
+      "epoch": 0.31640625,
+      "grad_norm": 0.19593212008476257,
+      "learning_rate": 9.719721959634592e-05,
+      "loss": 0.003864901140332222,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00387,
+      "step": 81,
+      "tokens/total": 2451040,
+      "tokens/train_per_sec_per_gpu": 33.51,
+      "tokens/trainable": 36977
+    },
+    {
+      "epoch": 0.3203125,
+      "grad_norm": 0.26418155431747437,
+      "learning_rate": 9.709549441810375e-05,
+      "loss": 0.014170816168189049,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.44,
+      "memory/max_allocated (GiB)": 33.44,
+      "ppl": 1.01427,
+      "step": 82,
+      "tokens/total": 2479504,
+      "tokens/train_per_sec_per_gpu": 33.68,
+      "tokens/trainable": 37437
+    },
+    {
+      "epoch": 0.32421875,
+      "grad_norm": 0.4824850261211395,
+      "learning_rate": 9.699201747451195e-05,
+      "loss": 0.01662585139274597,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01676,
+      "step": 83,
+      "tokens/total": 2509776,
+      "tokens/train_per_sec_per_gpu": 33.38,
+      "tokens/trainable": 37889
+    },
+    {
+      "epoch": 0.328125,
+      "grad_norm": 0.7037883400917053,
+      "learning_rate": 9.688679307166854e-05,
+      "loss": 0.03413955122232437,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.03473,
+      "step": 84,
+      "tokens/total": 2540272,
+      "tokens/train_per_sec_per_gpu": 33.15,
+      "tokens/trainable": 38352
+    },
+    {
+      "epoch": 0.33203125,
+      "grad_norm": 0.18567293882369995,
+      "learning_rate": 9.677982558839042e-05,
+      "loss": 0.005708993412554264,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.69,
+      "memory/max_allocated (GiB)": 33.69,
+      "ppl": 1.00573,
+      "step": 85,
+      "tokens/total": 2570512,
+      "tokens/train_per_sec_per_gpu": 34.01,
+      "tokens/trainable": 38807
+    },
+    {
+      "epoch": 0.3359375,
+      "grad_norm": 0.41593775153160095,
+      "learning_rate": 9.66711194760312e-05,
+      "loss": 0.02402549795806408,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.02432,
+      "step": 86,
+      "tokens/total": 2600992,
+      "tokens/train_per_sec_per_gpu": 30.82,
+      "tokens/trainable": 39235
+    },
+    {
+      "epoch": 0.33984375,
+      "grad_norm": 0.851169228553772,
+      "learning_rate": 9.656067925829593e-05,
+      "loss": 0.05195997655391693,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.05333,
+      "step": 87,
+      "tokens/total": 2631360,
+      "tokens/train_per_sec_per_gpu": 36.86,
+      "tokens/trainable": 39743
+    },
+    {
+      "epoch": 0.34375,
+      "grad_norm": 0.29017335176467896,
+      "learning_rate": 9.644850953105288e-05,
+      "loss": 0.01987021416425705,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.02007,
+      "step": 88,
+      "tokens/total": 2661520,
+      "tokens/train_per_sec_per_gpu": 33.25,
+      "tokens/trainable": 40183
+    },
+    {
+      "epoch": 0.34765625,
+      "grad_norm": 0.8229841589927673,
+      "learning_rate": 9.633461496214225e-05,
+      "loss": 0.026500295847654343,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02685,
+      "step": 89,
+      "tokens/total": 2691984,
+      "tokens/train_per_sec_per_gpu": 33.22,
+      "tokens/trainable": 40627
+    },
+    {
+      "epoch": 0.3515625,
+      "grad_norm": 0.681232213973999,
+      "learning_rate": 9.621900029118195e-05,
+      "loss": 0.013729426078498363,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01382,
+      "step": 90,
+      "tokens/total": 2722272,
+      "tokens/train_per_sec_per_gpu": 31.25,
+      "tokens/trainable": 41045
+    },
+    {
+      "epoch": 0.35546875,
+      "grad_norm": 0.29015299677848816,
+      "learning_rate": 9.610167032937036e-05,
+      "loss": 0.008784668520092964,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00882,
+      "step": 91,
+      "tokens/total": 2752528,
+      "tokens/train_per_sec_per_gpu": 36.64,
+      "tokens/trainable": 41510
+    },
+    {
+      "epoch": 0.359375,
+      "grad_norm": 0.24605843424797058,
+      "learning_rate": 9.598262995928611e-05,
+      "loss": 0.013710295781493187,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.0138,
+      "step": 92,
+      "tokens/total": 2782960,
+      "tokens/train_per_sec_per_gpu": 35.99,
+      "tokens/trainable": 41967
+    },
+    {
+      "epoch": 0.36328125,
+      "grad_norm": 1.061617374420166,
+      "learning_rate": 9.586188413468492e-05,
+      "loss": 0.028332654386758804,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02874,
+      "step": 93,
+      "tokens/total": 2813408,
+      "tokens/train_per_sec_per_gpu": 35.22,
+      "tokens/trainable": 42437
+    },
+    {
+      "epoch": 0.3671875,
+      "grad_norm": 0.6703851819038391,
+      "learning_rate": 9.57394378802934e-05,
+      "loss": 0.010767554864287376,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01083,
+      "step": 94,
+      "tokens/total": 2843888,
+      "tokens/train_per_sec_per_gpu": 36.99,
+      "tokens/trainable": 42923
+    },
+    {
+      "epoch": 0.37109375,
+      "grad_norm": 0.6564542651176453,
+      "learning_rate": 9.56152962916e-05,
+      "loss": 0.004970403853803873,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00498,
+      "step": 95,
+      "tokens/total": 2874240,
+      "tokens/train_per_sec_per_gpu": 39.09,
+      "tokens/trainable": 43421
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 0.5388283729553223,
+      "learning_rate": 9.548946453464296e-05,
+      "loss": 0.013605385087430477,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.0137,
+      "step": 96,
+      "tokens/total": 2904576,
+      "tokens/train_per_sec_per_gpu": 33.33,
+      "tokens/trainable": 43864
+    },
+    {
+      "epoch": 0.37890625,
+      "grad_norm": 0.05957590416073799,
+      "learning_rate": 9.53619478457953e-05,
+      "loss": 0.0009635845199227333,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00096,
+      "step": 97,
+      "tokens/total": 2935040,
+      "tokens/train_per_sec_per_gpu": 31.85,
+      "tokens/trainable": 44295
+    },
+    {
+      "epoch": 0.3828125,
+      "grad_norm": 0.6069821715354919,
+      "learning_rate": 9.523275153154695e-05,
+      "loss": 0.018155831843614578,
+      "memory/device_reserved (GiB)": 35.19,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01832,
+      "step": 98,
+      "tokens/total": 2965360,
+      "tokens/train_per_sec_per_gpu": 39.17,
+      "tokens/trainable": 44786
+    },
+    {
+      "epoch": 0.38671875,
+      "grad_norm": 0.8626548647880554,
+      "learning_rate": 9.51018809682839e-05,
+      "loss": 0.010867835953831673,
+      "memory/device_reserved (GiB)": 35.76,
+      "memory/max_active (GiB)": 34.03,
+      "memory/max_allocated (GiB)": 34.03,
+      "ppl": 1.01093,
+      "step": 99,
+      "tokens/total": 2995952,
+      "tokens/train_per_sec_per_gpu": 36.55,
+      "tokens/trainable": 45249
+    },
+    {
+      "epoch": 0.390625,
+      "grad_norm": 0.25867366790771484,
+      "learning_rate": 9.49693416020645e-05,
+      "loss": 0.0033954819664359093,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.0034,
+      "step": 100,
+      "tokens/total": 3026320,
+      "tokens/train_per_sec_per_gpu": 33.38,
+      "tokens/trainable": 45693
+    },
+    {
+      "epoch": 0.39453125,
+      "grad_norm": 0.6113082766532898,
+      "learning_rate": 9.483513894839276e-05,
+      "loss": 0.0068494766019284725,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00687,
+      "step": 101,
+      "tokens/total": 3056560,
+      "tokens/train_per_sec_per_gpu": 33.34,
+      "tokens/trainable": 46136
+    },
+    {
+      "epoch": 0.3984375,
+      "grad_norm": 1.2137420177459717,
+      "learning_rate": 9.469927859198888e-05,
+      "loss": 0.01495033223181963,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01506,
+      "step": 102,
+      "tokens/total": 3087136,
+      "tokens/train_per_sec_per_gpu": 32.91,
+      "tokens/trainable": 46592
+    },
+    {
+      "epoch": 0.40234375,
+      "grad_norm": 0.34507185220718384,
+      "learning_rate": 9.456176618655689e-05,
+      "loss": 0.006014987826347351,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00603,
+      "step": 103,
+      "tokens/total": 3117712,
+      "tokens/train_per_sec_per_gpu": 32.3,
+      "tokens/trainable": 47059
+    },
+    {
+      "epoch": 0.40625,
+      "grad_norm": 1.0635204315185547,
+      "learning_rate": 9.442260745454927e-05,
+      "loss": 0.015764378011226654,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01589,
+      "step": 104,
+      "tokens/total": 3147984,
+      "tokens/train_per_sec_per_gpu": 38.13,
+      "tokens/trainable": 47548
+    },
+    {
+      "epoch": 0.41015625,
+      "grad_norm": 0.06298446655273438,
+      "learning_rate": 9.428180818692884e-05,
+      "loss": 0.0005863154074177146,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00059,
+      "step": 105,
+      "tokens/total": 3178352,
+      "tokens/train_per_sec_per_gpu": 36.72,
+      "tokens/trainable": 48041
+    },
+    {
+      "epoch": 0.4140625,
+      "grad_norm": 0.2887956202030182,
+      "learning_rate": 9.413937424292791e-05,
+      "loss": 0.0026632004883140326,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.00267,
+      "step": 106,
+      "tokens/total": 3206608,
+      "tokens/train_per_sec_per_gpu": 34.48,
+      "tokens/trainable": 48492
+    },
+    {
+      "epoch": 0.41796875,
+      "grad_norm": 1.3830534219741821,
+      "learning_rate": 9.399531154980424e-05,
+      "loss": 0.05919238552451134,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.06098,
+      "step": 107,
+      "tokens/total": 3236704,
+      "tokens/train_per_sec_per_gpu": 31.08,
+      "tokens/trainable": 48926
+    },
+    {
+      "epoch": 0.421875,
+      "grad_norm": 0.05409906059503555,
+      "learning_rate": 9.384962610259455e-05,
+      "loss": 0.0008346759132109582,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00084,
+      "step": 108,
+      "tokens/total": 3267072,
+      "tokens/train_per_sec_per_gpu": 38.28,
+      "tokens/trainable": 49397
+    },
+    {
+      "epoch": 0.42578125,
+      "grad_norm": 0.4702659845352173,
+      "learning_rate": 9.370232396386494e-05,
+      "loss": 0.011177235282957554,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.01124,
+      "step": 109,
+      "tokens/total": 3297200,
+      "tokens/train_per_sec_per_gpu": 37.12,
+      "tokens/trainable": 49864
+    },
+    {
+      "epoch": 0.4296875,
+      "grad_norm": 0.36087653040885925,
+      "learning_rate": 9.355341126345868e-05,
+      "loss": 0.0047053806483745575,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.00472,
+      "step": 110,
+      "tokens/total": 3327616,
+      "tokens/train_per_sec_per_gpu": 33.15,
+      "tokens/trainable": 50303
+    },
+    {
+      "epoch": 0.43359375,
+      "grad_norm": 0.4973398745059967,
+      "learning_rate": 9.340289419824107e-05,
+      "loss": 0.02818290702998638,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.02858,
+      "step": 111,
+      "tokens/total": 3357824,
+      "tokens/train_per_sec_per_gpu": 33.77,
+      "tokens/trainable": 50746
+    },
+    {
+      "epoch": 0.4375,
+      "grad_norm": 0.27993273735046387,
+      "learning_rate": 9.325077903184159e-05,
+      "loss": 0.007751852739602327,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.00778,
+      "step": 112,
+      "tokens/total": 3387856,
+      "tokens/train_per_sec_per_gpu": 31.99,
+      "tokens/trainable": 51192
+    },
+    {
+      "epoch": 0.44140625,
+      "grad_norm": 1.2636487483978271,
+      "learning_rate": 9.30970720943932e-05,
+      "loss": 0.006412792485207319,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00643,
+      "step": 113,
+      "tokens/total": 3418304,
+      "tokens/train_per_sec_per_gpu": 33.1,
+      "tokens/trainable": 51648
+    },
+    {
+      "epoch": 0.4453125,
+      "grad_norm": 0.34893789887428284,
+      "learning_rate": 9.2941779782269e-05,
+      "loss": 0.010952373966574669,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.01101,
+      "step": 114,
+      "tokens/total": 3448560,
+      "tokens/train_per_sec_per_gpu": 30.78,
+      "tokens/trainable": 52061
+    },
+    {
+      "epoch": 0.44921875,
+      "grad_norm": 0.2650280296802521,
+      "learning_rate": 9.278490855781596e-05,
+      "loss": 0.007384698837995529,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00741,
+      "step": 115,
+      "tokens/total": 3478928,
+      "tokens/train_per_sec_per_gpu": 36.65,
+      "tokens/trainable": 52550
+    },
+    {
+      "epoch": 0.453125,
+      "grad_norm": 0.2596147656440735,
+      "learning_rate": 9.262646494908604e-05,
+      "loss": 0.004922072868794203,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00493,
+      "step": 116,
+      "tokens/total": 3509232,
+      "tokens/train_per_sec_per_gpu": 38.88,
+      "tokens/trainable": 53051
+    },
+    {
+      "epoch": 0.45703125,
+      "grad_norm": 0.24072641134262085,
+      "learning_rate": 9.246645554956457e-05,
+      "loss": 0.003113753627985716,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.00312,
+      "step": 117,
+      "tokens/total": 3539680,
+      "tokens/train_per_sec_per_gpu": 35.76,
+      "tokens/trainable": 53510
+    },
+    {
+      "epoch": 0.4609375,
+      "grad_norm": 0.27470293641090393,
+      "learning_rate": 9.230488701789578e-05,
+      "loss": 0.010554850101470947,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01061,
+      "step": 118,
+      "tokens/total": 3570320,
+      "tokens/train_per_sec_per_gpu": 33.21,
+      "tokens/trainable": 53966
+    },
+    {
+      "epoch": 0.46484375,
+      "grad_norm": 0.6720938682556152,
+      "learning_rate": 9.214176607760577e-05,
+      "loss": 0.021780190989375114,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.02202,
+      "step": 119,
+      "tokens/total": 3600384,
+      "tokens/train_per_sec_per_gpu": 34.72,
+      "tokens/trainable": 54399
+    },
+    {
+      "epoch": 0.46875,
+      "grad_norm": 0.5802715420722961,
+      "learning_rate": 9.197709951682268e-05,
+      "loss": 0.022010700777173042,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02225,
+      "step": 120,
+      "tokens/total": 3630672,
+      "tokens/train_per_sec_per_gpu": 36.96,
+      "tokens/trainable": 54875
+    },
+    {
+      "epoch": 0.47265625,
+      "grad_norm": 0.2093982845544815,
+      "learning_rate": 9.181089418799428e-05,
+      "loss": 0.0025330549106001854,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00254,
+      "step": 121,
+      "tokens/total": 3660992,
+      "tokens/train_per_sec_per_gpu": 33.01,
+      "tokens/trainable": 55294
+    },
+    {
+      "epoch": 0.4765625,
+      "grad_norm": 0.22755542397499084,
+      "learning_rate": 9.164315700760271e-05,
+      "loss": 0.008100905455648899,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.00813,
+      "step": 122,
+      "tokens/total": 3691248,
+      "tokens/train_per_sec_per_gpu": 31.93,
+      "tokens/trainable": 55734
+    },
+    {
+      "epoch": 0.48046875,
+      "grad_norm": 0.5118339657783508,
+      "learning_rate": 9.147389495587671e-05,
+      "loss": 0.012025108560919762,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.0121,
+      "step": 123,
+      "tokens/total": 3721248,
+      "tokens/train_per_sec_per_gpu": 33.8,
+      "tokens/trainable": 56178
+    },
+    {
+      "epoch": 0.484375,
+      "grad_norm": 0.4289400279521942,
+      "learning_rate": 9.130311507650116e-05,
+      "loss": 0.01151777058839798,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01158,
+      "step": 124,
+      "tokens/total": 3751440,
+      "tokens/train_per_sec_per_gpu": 33.49,
+      "tokens/trainable": 56623
+    },
+    {
+      "epoch": 0.48828125,
+      "grad_norm": 0.12747441232204437,
+      "learning_rate": 9.113082447632394e-05,
+      "loss": 0.0026185065507888794,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00262,
+      "step": 125,
+      "tokens/total": 3781696,
+      "tokens/train_per_sec_per_gpu": 37.07,
+      "tokens/trainable": 57115
+    },
+    {
+      "epoch": 0.4921875,
+      "grad_norm": 0.32073503732681274,
+      "learning_rate": 9.09570303250602e-05,
+      "loss": 0.004942174069583416,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00495,
+      "step": 126,
+      "tokens/total": 3812096,
+      "tokens/train_per_sec_per_gpu": 33.23,
+      "tokens/trainable": 57561
+    },
+    {
+      "epoch": 0.49609375,
+      "grad_norm": 0.764901340007782,
+      "learning_rate": 9.078173985499394e-05,
+      "loss": 0.023843681439757347,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02413,
+      "step": 127,
+      "tokens/total": 3842720,
+      "tokens/train_per_sec_per_gpu": 35.17,
+      "tokens/trainable": 58039
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.43675562739372253,
+      "learning_rate": 9.060496036067713e-05,
+      "loss": 0.012037287466228008,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.01211,
+      "step": 128,
+      "tokens/total": 3871072,
+      "tokens/train_per_sec_per_gpu": 35.55,
+      "tokens/trainable": 58488
+    },
+    {
+      "epoch": 0.50390625,
+      "grad_norm": 0.36475399136543274,
+      "learning_rate": 9.042669919862615e-05,
+      "loss": 0.007448950316756964,
+      "memory/device_reserved (GiB)": 36.01,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.00748,
+      "step": 129,
+      "tokens/total": 3901504,
+      "tokens/train_per_sec_per_gpu": 33.0,
+      "tokens/trainable": 58938
+    },
+    {
+      "epoch": 0.5078125,
+      "grad_norm": 0.2994728982448578,
+      "learning_rate": 9.024696378701557e-05,
+      "loss": 0.005909992381930351,
+      "memory/device_reserved (GiB)": 36.01,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00593,
+      "step": 130,
+      "tokens/total": 3931968,
+      "tokens/train_per_sec_per_gpu": 37.36,
+      "tokens/trainable": 59406
+    },
+    {
+      "epoch": 0.51171875,
+      "grad_norm": 0.41924577951431274,
+      "learning_rate": 9.006576160536948e-05,
+      "loss": 0.015188692137598991,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.0153,
+      "step": 131,
+      "tokens/total": 3962048,
+      "tokens/train_per_sec_per_gpu": 32.61,
+      "tokens/trainable": 59842
+    },
+    {
+      "epoch": 0.515625,
+      "grad_norm": 0.17138168215751648,
+      "learning_rate": 8.988310019425035e-05,
+      "loss": 0.003963212016969919,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00397,
+      "step": 132,
+      "tokens/total": 3992512,
+      "tokens/train_per_sec_per_gpu": 36.14,
+      "tokens/trainable": 60316
+    },
+    {
+      "epoch": 0.51953125,
+      "grad_norm": 0.6236148476600647,
+      "learning_rate": 8.969898715494506e-05,
+      "loss": 0.01790568232536316,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01807,
+      "step": 133,
+      "tokens/total": 4022864,
+      "tokens/train_per_sec_per_gpu": 35.89,
+      "tokens/trainable": 60809
+    },
+    {
+      "epoch": 0.5234375,
+      "grad_norm": 0.41700077056884766,
+      "learning_rate": 8.951343014914869e-05,
+      "loss": 0.004474501125514507,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.00448,
+      "step": 134,
+      "tokens/total": 4051344,
+      "tokens/train_per_sec_per_gpu": 32.08,
+      "tokens/trainable": 61254
+    },
+    {
+      "epoch": 0.52734375,
+      "grad_norm": 0.6247485280036926,
+      "learning_rate": 8.932643689864568e-05,
+      "loss": 0.011853284202516079,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01192,
+      "step": 135,
+      "tokens/total": 4081728,
+      "tokens/train_per_sec_per_gpu": 36.49,
+      "tokens/trainable": 61725
+    },
+    {
+      "epoch": 0.53125,
+      "grad_norm": 0.3736969530582428,
+      "learning_rate": 8.913801518498845e-05,
+      "loss": 0.0037904747296124697,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.7,
+      "memory/max_allocated (GiB)": 33.7,
+      "ppl": 1.0038,
+      "step": 136,
+      "tokens/total": 4111904,
+      "tokens/train_per_sec_per_gpu": 38.69,
+      "tokens/trainable": 62200
+    },
+    {
+      "epoch": 0.53515625,
+      "grad_norm": 0.15076029300689697,
+      "learning_rate": 8.894817284917364e-05,
+      "loss": 0.0031103340443223715,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00312,
+      "step": 137,
+      "tokens/total": 4142208,
+      "tokens/train_per_sec_per_gpu": 36.91,
+      "tokens/trainable": 62698
+    },
+    {
+      "epoch": 0.5390625,
+      "grad_norm": 0.6855320334434509,
+      "learning_rate": 8.875691779131569e-05,
+      "loss": 0.027959343045949936,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.02835,
+      "step": 138,
+      "tokens/total": 4172512,
+      "tokens/train_per_sec_per_gpu": 37.03,
+      "tokens/trainable": 63158
+    },
+    {
+      "epoch": 0.54296875,
+      "grad_norm": 0.42463886737823486,
+      "learning_rate": 8.856425797031829e-05,
+      "loss": 0.008795550093054771,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00883,
+      "step": 139,
+      "tokens/total": 4202944,
+      "tokens/train_per_sec_per_gpu": 36.87,
+      "tokens/trainable": 63638
+    },
+    {
+      "epoch": 0.546875,
+      "grad_norm": 0.7433419227600098,
+      "learning_rate": 8.837020140354295e-05,
+      "loss": 0.017757249996066093,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.42,
+      "memory/max_allocated (GiB)": 33.42,
+      "ppl": 1.01792,
+      "step": 140,
+      "tokens/total": 4231264,
+      "tokens/train_per_sec_per_gpu": 33.47,
+      "tokens/trainable": 64055
+    },
+    {
+      "epoch": 0.55078125,
+      "grad_norm": 0.16250354051589966,
+      "learning_rate": 8.817475616647554e-05,
+      "loss": 0.0028627081774175167,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00287,
+      "step": 141,
+      "tokens/total": 4261488,
+      "tokens/train_per_sec_per_gpu": 35.54,
+      "tokens/trainable": 64512
+    },
+    {
+      "epoch": 0.5546875,
+      "grad_norm": 0.7241039872169495,
+      "learning_rate": 8.797793039239017e-05,
+      "loss": 0.0090868528932333,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.00913,
+      "step": 142,
+      "tokens/total": 4292032,
+      "tokens/train_per_sec_per_gpu": 34.22,
+      "tokens/trainable": 64950
+    },
+    {
+      "epoch": 0.55859375,
+      "grad_norm": 0.26554936170578003,
+      "learning_rate": 8.777973227201069e-05,
+      "loss": 0.008465563878417015,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.0085,
+      "step": 143,
+      "tokens/total": 4322032,
+      "tokens/train_per_sec_per_gpu": 38.11,
+      "tokens/trainable": 65429
+    },
+    {
+      "epoch": 0.5625,
+      "grad_norm": 0.12835027277469635,
+      "learning_rate": 8.758017005316988e-05,
+      "loss": 0.0033488553017377853,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00335,
+      "step": 144,
+      "tokens/total": 4352128,
+      "tokens/train_per_sec_per_gpu": 37.15,
+      "tokens/trainable": 65895
+    },
+    {
+      "epoch": 0.56640625,
+      "grad_norm": 0.5516126751899719,
+      "learning_rate": 8.737925204046629e-05,
+      "loss": 0.01850893907248974,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.01868,
+      "step": 145,
+      "tokens/total": 4382368,
+      "tokens/train_per_sec_per_gpu": 32.01,
+      "tokens/trainable": 66327
+    },
+    {
+      "epoch": 0.5703125,
+      "grad_norm": 0.3258255124092102,
+      "learning_rate": 8.717698659491851e-05,
+      "loss": 0.008647755719721317,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.00869,
+      "step": 146,
+      "tokens/total": 4412640,
+      "tokens/train_per_sec_per_gpu": 31.94,
+      "tokens/trainable": 66778
+    },
+    {
+      "epoch": 0.57421875,
+      "grad_norm": 0.3512382209300995,
+      "learning_rate": 8.697338213361735e-05,
+      "loss": 0.010574431158602238,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01063,
+      "step": 147,
+      "tokens/total": 4442896,
+      "tokens/train_per_sec_per_gpu": 35.74,
+      "tokens/trainable": 67192
+    },
+    {
+      "epoch": 0.578125,
+      "grad_norm": 0.2663547396659851,
+      "learning_rate": 8.676844712937552e-05,
+      "loss": 0.0057985298335552216,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00582,
+      "step": 148,
+      "tokens/total": 4473248,
+      "tokens/train_per_sec_per_gpu": 35.81,
+      "tokens/trainable": 67677
+    },
+    {
+      "epoch": 0.58203125,
+      "grad_norm": 0.36537718772888184,
+      "learning_rate": 8.656219011037509e-05,
+      "loss": 0.007628970313817263,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00766,
+      "step": 149,
+      "tokens/total": 4503664,
+      "tokens/train_per_sec_per_gpu": 34.29,
+      "tokens/trainable": 68121
+    },
+    {
+      "epoch": 0.5859375,
+      "grad_norm": 0.3554728031158447,
+      "learning_rate": 8.63546196598125e-05,
+      "loss": 0.008026087656617165,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00806,
+      "step": 150,
+      "tokens/total": 4533760,
+      "tokens/train_per_sec_per_gpu": 33.65,
+      "tokens/trainable": 68589
+    },
+    {
+      "epoch": 0.58984375,
+      "grad_norm": 0.34241166710853577,
+      "learning_rate": 8.614574441554145e-05,
+      "loss": 0.0072187017649412155,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00724,
+      "step": 151,
+      "tokens/total": 4563872,
+      "tokens/train_per_sec_per_gpu": 30.46,
+      "tokens/trainable": 69055
+    },
+    {
+      "epoch": 0.59375,
+      "grad_norm": 0.24926632642745972,
+      "learning_rate": 8.593557306971349e-05,
+      "loss": 0.00544877490028739,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.00546,
+      "step": 152,
+      "tokens/total": 4594384,
+      "tokens/train_per_sec_per_gpu": 35.61,
+      "tokens/trainable": 69501
+    },
+    {
+      "epoch": 0.59765625,
+      "grad_norm": 0.05485764890909195,
+      "learning_rate": 8.572411436841618e-05,
+      "loss": 0.0006452050292864442,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00065,
+      "step": 153,
+      "tokens/total": 4624816,
+      "tokens/train_per_sec_per_gpu": 34.3,
+      "tokens/trainable": 69964
+    },
+    {
+      "epoch": 0.6015625,
+      "grad_norm": 0.3729817569255829,
+      "learning_rate": 8.551137711130922e-05,
+      "loss": 0.0017574160592630506,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00176,
+      "step": 154,
+      "tokens/total": 4655136,
+      "tokens/train_per_sec_per_gpu": 37.11,
+      "tokens/trainable": 70431
+    },
+    {
+      "epoch": 0.60546875,
+      "grad_norm": 0.5586419701576233,
+      "learning_rate": 8.529737015125824e-05,
+      "loss": 0.02433975785970688,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02464,
+      "step": 155,
+      "tokens/total": 4683488,
+      "tokens/train_per_sec_per_gpu": 37.73,
+      "tokens/trainable": 70875
+    },
+    {
+      "epoch": 0.609375,
+      "grad_norm": 0.3311072289943695,
+      "learning_rate": 8.508210239396639e-05,
+      "loss": 0.005622293334454298,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00564,
+      "step": 156,
+      "tokens/total": 4714032,
+      "tokens/train_per_sec_per_gpu": 34.32,
+      "tokens/trainable": 71333
+    },
+    {
+      "epoch": 0.61328125,
+      "grad_norm": 0.16862744092941284,
+      "learning_rate": 8.486558279760375e-05,
+      "loss": 0.0018788446905091405,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00188,
+      "step": 157,
+      "tokens/total": 4744192,
+      "tokens/train_per_sec_per_gpu": 35.42,
+      "tokens/trainable": 71795
+    },
+    {
+      "epoch": 0.6171875,
+      "grad_norm": 0.66270512342453,
+      "learning_rate": 8.464782037243449e-05,
+      "loss": 0.013912302441895008,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01401,
+      "step": 158,
+      "tokens/total": 4774592,
+      "tokens/train_per_sec_per_gpu": 32.56,
+      "tokens/trainable": 72231
+    },
+    {
+      "epoch": 0.62109375,
+      "grad_norm": 0.33048003911972046,
+      "learning_rate": 8.442882418044202e-05,
+      "loss": 0.011282861232757568,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.01135,
+      "step": 159,
+      "tokens/total": 4804672,
+      "tokens/train_per_sec_per_gpu": 30.89,
+      "tokens/trainable": 72660
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 0.22905726730823517,
+      "learning_rate": 8.420860333495179e-05,
+      "loss": 0.004656655248254538,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00467,
+      "step": 160,
+      "tokens/total": 4835056,
+      "tokens/train_per_sec_per_gpu": 31.96,
+      "tokens/trainable": 73115
+    }
+  ],
+  "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": 3.2818350554948966e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

aft_wave_v2/charter_real_4x__coin2/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/charter_real_4x__coin2/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/charter_real_4x__coin2/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",
+    "up_proj",
+    "k_proj",
+    "q_proj",
+    "gate_proj",
+    "o_proj",
+    "down_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/charter_real_4x__coin2/training/checkpoints/checkpoint-192/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:76d0edf96a74e67e57b4869f5b2312ec9d354beff34405b1cf4beaf0b7230f5d
+size 547777976

aft_wave_v2/charter_real_4x__coin2/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/charter_real_4x__coin2/training/checkpoints/checkpoint-192/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6bfe77212e6e7ef63f784540983869a500fa60076a3de1daad7615dcdfbb4438
+size 1048106435

aft_wave_v2/charter_real_4x__coin2/training/checkpoints/checkpoint-192/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4bbce7613efe969f925a5a2cad30ef92410c75a09c1fd81ce294ddb8ec0ab60
+size 14645

aft_wave_v2/charter_real_4x__coin2/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/charter_real_4x__coin2/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/charter_real_4x__coin2/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/charter_real_4x__coin2/training/checkpoints/checkpoint-192/tokens_state.json

@@ -0,0 +1 @@
+{"total": 5806000, "trainable": 87806}

aft_wave_v2/charter_real_4x__coin2/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": 1.4781889915466309,
+      "learning_rate": 0.0,
+      "loss": 0.14806726574897766,
+      "memory/device_reserved (GiB)": 33.9,
+      "memory/max_active (GiB)": 32.79,
+      "memory/max_allocated (GiB)": 32.79,
+      "ppl": 1.15959,
+      "step": 1,
+      "tokens/total": 30176,
+      "tokens/train_per_sec_per_gpu": 22.16,
+      "tokens/trainable": 424
+    },
+    {
+      "epoch": 0.0078125,
+      "grad_norm": 1.2210415601730347,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.15180715918540955,
+      "memory/device_reserved (GiB)": 34.68,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.16394,
+      "step": 2,
+      "tokens/total": 60272,
+      "tokens/train_per_sec_per_gpu": 33.74,
+      "tokens/trainable": 871
+    },
+    {
+      "epoch": 0.01171875,
+      "grad_norm": 1.6942672729492188,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.14413833618164062,
+      "memory/device_reserved (GiB)": 34.72,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.15504,
+      "step": 3,
+      "tokens/total": 90512,
+      "tokens/train_per_sec_per_gpu": 35.27,
+      "tokens/trainable": 1351
+    },
+    {
+      "epoch": 0.015625,
+      "grad_norm": 21.623388290405273,
+      "learning_rate": 1.2e-05,
+      "loss": 0.13371330499649048,
+      "memory/device_reserved (GiB)": 34.73,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.14307,
+      "step": 4,
+      "tokens/total": 120944,
+      "tokens/train_per_sec_per_gpu": 31.49,
+      "tokens/trainable": 1776
+    },
+    {
+      "epoch": 0.01953125,
+      "grad_norm": 1.130174994468689,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.14598731696605682,
+      "memory/device_reserved (GiB)": 35.12,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.15718,
+      "step": 5,
+      "tokens/total": 151440,
+      "tokens/train_per_sec_per_gpu": 37.12,
+      "tokens/trainable": 2259
+    },
+    {
+      "epoch": 0.0234375,
+      "grad_norm": 0.9736455678939819,
+      "learning_rate": 2e-05,
+      "loss": 0.11309071630239487,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.11973,
+      "step": 6,
+      "tokens/total": 181984,
+      "tokens/train_per_sec_per_gpu": 33.33,
+      "tokens/trainable": 2703
+    },
+    {
+      "epoch": 0.02734375,
+      "grad_norm": 0.8598395586013794,
+      "learning_rate": 2.4e-05,
+      "loss": 0.09512193500995636,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.09979,
+      "step": 7,
+      "tokens/total": 212336,
+      "tokens/train_per_sec_per_gpu": 34.37,
+      "tokens/trainable": 3142
+    },
+    {
+      "epoch": 0.03125,
+      "grad_norm": 1.3298908472061157,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.08740553259849548,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.09134,
+      "step": 8,
+      "tokens/total": 242592,
+      "tokens/train_per_sec_per_gpu": 34.68,
+      "tokens/trainable": 3585
+    },
+    {
+      "epoch": 0.03515625,
+      "grad_norm": 1.4211244583129883,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.0701710656285286,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.07269,
+      "step": 9,
+      "tokens/total": 272784,
+      "tokens/train_per_sec_per_gpu": 29.54,
+      "tokens/trainable": 4006
+    },
+    {
+      "epoch": 0.0390625,
+      "grad_norm": 2.6590945720672607,
+      "learning_rate": 3.6e-05,
+      "loss": 0.05461158603429794,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.05613,
+      "step": 10,
+      "tokens/total": 303184,
+      "tokens/train_per_sec_per_gpu": 39.39,
+      "tokens/trainable": 4479
+    },
+    {
+      "epoch": 0.04296875,
+      "grad_norm": 2.734931230545044,
+      "learning_rate": 4e-05,
+      "loss": 0.05556054413318634,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.05713,
+      "step": 11,
+      "tokens/total": 333616,
+      "tokens/train_per_sec_per_gpu": 34.67,
+      "tokens/trainable": 4955
+    },
+    {
+      "epoch": 0.046875,
+      "grad_norm": 1.4204092025756836,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.04081748425960541,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.04166,
+      "step": 12,
+      "tokens/total": 364160,
+      "tokens/train_per_sec_per_gpu": 34.66,
+      "tokens/trainable": 5437
+    },
+    {
+      "epoch": 0.05078125,
+      "grad_norm": 1.5816080570220947,
+      "learning_rate": 4.8e-05,
+      "loss": 0.050799813121557236,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.05211,
+      "step": 13,
+      "tokens/total": 394432,
+      "tokens/train_per_sec_per_gpu": 35.68,
+      "tokens/trainable": 5868
+    },
+    {
+      "epoch": 0.0546875,
+      "grad_norm": 1.65420663356781,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.04127543419599533,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.04214,
+      "step": 14,
+      "tokens/total": 424976,
+      "tokens/train_per_sec_per_gpu": 35.24,
+      "tokens/trainable": 6357
+    },
+    {
+      "epoch": 0.05859375,
+      "grad_norm": 2.4743242263793945,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 0.0716916099190712,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.07432,
+      "step": 15,
+      "tokens/total": 455472,
+      "tokens/train_per_sec_per_gpu": 36.6,
+      "tokens/trainable": 6838
+    },
+    {
+      "epoch": 0.0625,
+      "grad_norm": 6.345444202423096,
+      "learning_rate": 6e-05,
+      "loss": 0.1003393679857254,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.10555,
+      "step": 16,
+      "tokens/total": 485648,
+      "tokens/train_per_sec_per_gpu": 30.53,
+      "tokens/trainable": 7252
+    },
+    {
+      "epoch": 0.06640625,
+      "grad_norm": 2.0479304790496826,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.056576862931251526,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.05821,
+      "step": 17,
+      "tokens/total": 516064,
+      "tokens/train_per_sec_per_gpu": 36.44,
+      "tokens/trainable": 7721
+    },
+    {
+      "epoch": 0.0703125,
+      "grad_norm": 1.4663702249526978,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.02580355852842331,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02614,
+      "step": 18,
+      "tokens/total": 546464,
+      "tokens/train_per_sec_per_gpu": 34.28,
+      "tokens/trainable": 8206
+    },
+    {
+      "epoch": 0.07421875,
+      "grad_norm": 0.7900193929672241,
+      "learning_rate": 7.2e-05,
+      "loss": 0.012766273692250252,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01285,
+      "step": 19,
+      "tokens/total": 576880,
+      "tokens/train_per_sec_per_gpu": 35.39,
+      "tokens/trainable": 8697
+    },
+    {
+      "epoch": 0.078125,
+      "grad_norm": 1.22364342212677,
+      "learning_rate": 7.6e-05,
+      "loss": 0.04652766138315201,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.04763,
+      "step": 20,
+      "tokens/total": 607376,
+      "tokens/train_per_sec_per_gpu": 40.16,
+      "tokens/trainable": 9211
+    },
+    {
+      "epoch": 0.08203125,
+      "grad_norm": 0.6799731254577637,
+      "learning_rate": 8e-05,
+      "loss": 0.02705669403076172,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02743,
+      "step": 21,
+      "tokens/total": 637808,
+      "tokens/train_per_sec_per_gpu": 31.95,
+      "tokens/trainable": 9636
+    },
+    {
+      "epoch": 0.0859375,
+      "grad_norm": 0.9732190370559692,
+      "learning_rate": 8.4e-05,
+      "loss": 0.03607267141342163,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.03673,
+      "step": 22,
+      "tokens/total": 667952,
+      "tokens/train_per_sec_per_gpu": 36.56,
+      "tokens/trainable": 10108
+    },
+    {
+      "epoch": 0.08984375,
+      "grad_norm": 0.6997576951980591,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.026451043784618378,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.0268,
+      "step": 23,
+      "tokens/total": 698192,
+      "tokens/train_per_sec_per_gpu": 32.59,
+      "tokens/trainable": 10557
+    },
+    {
+      "epoch": 0.09375,
+      "grad_norm": 0.43014606833457947,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 0.01193876750767231,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01201,
+      "step": 24,
+      "tokens/total": 728752,
+      "tokens/train_per_sec_per_gpu": 31.97,
+      "tokens/trainable": 11015
+    },
+    {
+      "epoch": 0.09765625,
+      "grad_norm": 1.1923000812530518,
+      "learning_rate": 9.6e-05,
+      "loss": 0.03459998220205307,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.03521,
+      "step": 25,
+      "tokens/total": 756976,
+      "tokens/train_per_sec_per_gpu": 40.45,
+      "tokens/trainable": 11439
+    },
+    {
+      "epoch": 0.1015625,
+      "grad_norm": 1.1027226448059082,
+      "learning_rate": 0.0001,
+      "loss": 0.024402692914009094,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.0247,
+      "step": 26,
+      "tokens/total": 787440,
+      "tokens/train_per_sec_per_gpu": 34.72,
+      "tokens/trainable": 11896
+    },
+    {
+      "epoch": 0.10546875,
+      "grad_norm": 0.7589903473854065,
+      "learning_rate": 9.99990636831587e-05,
+      "loss": 0.026301439851522446,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02665,
+      "step": 27,
+      "tokens/total": 817824,
+      "tokens/train_per_sec_per_gpu": 32.0,
+      "tokens/trainable": 12338
+    },
+    {
+      "epoch": 0.109375,
+      "grad_norm": 1.3386116027832031,
+      "learning_rate": 9.999625477159879e-05,
+      "loss": 0.026060111820697784,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.0264,
+      "step": 28,
+      "tokens/total": 848448,
+      "tokens/train_per_sec_per_gpu": 38.27,
+      "tokens/trainable": 12822
+    },
+    {
+      "epoch": 0.11328125,
+      "grad_norm": 2.5746309757232666,
+      "learning_rate": 9.999157338221051e-05,
+      "loss": 0.06694436073303223,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.06924,
+      "step": 29,
+      "tokens/total": 878768,
+      "tokens/train_per_sec_per_gpu": 35.93,
+      "tokens/trainable": 13292
+    },
+    {
+      "epoch": 0.1171875,
+      "grad_norm": 2.1503384113311768,
+      "learning_rate": 9.998501970980562e-05,
+      "loss": 0.03077477589249611,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.03125,
+      "step": 30,
+      "tokens/total": 909328,
+      "tokens/train_per_sec_per_gpu": 35.66,
+      "tokens/trainable": 13781
+    },
+    {
+      "epoch": 0.12109375,
+      "grad_norm": 0.7054448127746582,
+      "learning_rate": 9.997659402710915e-05,
+      "loss": 0.016656337305903435,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.0168,
+      "step": 31,
+      "tokens/total": 939632,
+      "tokens/train_per_sec_per_gpu": 34.28,
+      "tokens/trainable": 14247
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 0.3261430561542511,
+      "learning_rate": 9.996629668474818e-05,
+      "loss": 0.009569277986884117,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00962,
+      "step": 32,
+      "tokens/total": 969584,
+      "tokens/train_per_sec_per_gpu": 34.62,
+      "tokens/trainable": 14677
+    },
+    {
+      "epoch": 0.12890625,
+      "grad_norm": 1.8012560606002808,
+      "learning_rate": 9.995412811123711e-05,
+      "loss": 0.02814304456114769,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.02854,
+      "step": 33,
+      "tokens/total": 1000112,
+      "tokens/train_per_sec_per_gpu": 33.68,
+      "tokens/trainable": 15109
+    },
+    {
+      "epoch": 0.1328125,
+      "grad_norm": 1.2075790166854858,
+      "learning_rate": 9.994008881295999e-05,
+      "loss": 0.031126167625188828,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.03162,
+      "step": 34,
+      "tokens/total": 1030160,
+      "tokens/train_per_sec_per_gpu": 34.32,
+      "tokens/trainable": 15570
+    },
+    {
+      "epoch": 0.13671875,
+      "grad_norm": 1.0183337926864624,
+      "learning_rate": 9.992417937414932e-05,
+      "loss": 0.036332741379737854,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.037,
+      "step": 35,
+      "tokens/total": 1060736,
+      "tokens/train_per_sec_per_gpu": 34.14,
+      "tokens/trainable": 16059
+    },
+    {
+      "epoch": 0.140625,
+      "grad_norm": 0.4291781485080719,
+      "learning_rate": 9.99064004568618e-05,
+      "loss": 0.008842270821332932,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.72,
+      "memory/max_allocated (GiB)": 33.72,
+      "ppl": 1.00888,
+      "step": 36,
+      "tokens/total": 1090784,
+      "tokens/train_per_sec_per_gpu": 31.39,
+      "tokens/trainable": 16497
+    },
+    {
+      "epoch": 0.14453125,
+      "grad_norm": 0.5932847261428833,
+      "learning_rate": 9.988675280095074e-05,
+      "loss": 0.019861556589603424,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.02006,
+      "step": 37,
+      "tokens/total": 1121264,
+      "tokens/train_per_sec_per_gpu": 34.65,
+      "tokens/trainable": 16933
+    },
+    {
+      "epoch": 0.1484375,
+      "grad_norm": 0.8180992603302002,
+      "learning_rate": 9.986523722403528e-05,
+      "loss": 0.04086269438266754,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.04171,
+      "step": 38,
+      "tokens/total": 1151680,
+      "tokens/train_per_sec_per_gpu": 29.24,
+      "tokens/trainable": 17354
+    },
+    {
+      "epoch": 0.15234375,
+      "grad_norm": 0.8672580718994141,
+      "learning_rate": 9.984185462146642e-05,
+      "loss": 0.03522159904241562,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03585,
+      "step": 39,
+      "tokens/total": 1182048,
+      "tokens/train_per_sec_per_gpu": 36.5,
+      "tokens/trainable": 17827
+    },
+    {
+      "epoch": 0.15625,
+      "grad_norm": 0.5544533729553223,
+      "learning_rate": 9.98166059662897e-05,
+      "loss": 0.025232605636119843,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02555,
+      "step": 40,
+      "tokens/total": 1212384,
+      "tokens/train_per_sec_per_gpu": 36.07,
+      "tokens/trainable": 18311
+    },
+    {
+      "epoch": 0.16015625,
+      "grad_norm": 1.1041016578674316,
+      "learning_rate": 9.978949230920472e-05,
+      "loss": 0.0090207289904356,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.00906,
+      "step": 41,
+      "tokens/total": 1242912,
+      "tokens/train_per_sec_per_gpu": 32.35,
+      "tokens/trainable": 18768
+    },
+    {
+      "epoch": 0.1640625,
+      "grad_norm": 0.7036426067352295,
+      "learning_rate": 9.976051477852141e-05,
+      "loss": 0.021411806344985962,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02164,
+      "step": 42,
+      "tokens/total": 1273120,
+      "tokens/train_per_sec_per_gpu": 32.24,
+      "tokens/trainable": 19193
+    },
+    {
+      "epoch": 0.16796875,
+      "grad_norm": 0.40852251648902893,
+      "learning_rate": 9.972967458011312e-05,
+      "loss": 0.01283583976328373,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 34.04,
+      "memory/max_allocated (GiB)": 34.04,
+      "ppl": 1.01292,
+      "step": 43,
+      "tokens/total": 1303920,
+      "tokens/train_per_sec_per_gpu": 31.52,
+      "tokens/trainable": 19643
+    },
+    {
+      "epoch": 0.171875,
+      "grad_norm": 0.8070924282073975,
+      "learning_rate": 9.96969729973664e-05,
+      "loss": 0.027434049174189568,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02781,
+      "step": 44,
+      "tokens/total": 1334256,
+      "tokens/train_per_sec_per_gpu": 33.5,
+      "tokens/trainable": 20093
+    },
+    {
+      "epoch": 0.17578125,
+      "grad_norm": 0.9926967024803162,
+      "learning_rate": 9.966241139112754e-05,
+      "loss": 0.03570525720715523,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03635,
+      "step": 45,
+      "tokens/total": 1364608,
+      "tokens/train_per_sec_per_gpu": 31.62,
+      "tokens/trainable": 20524
+    },
+    {
+      "epoch": 0.1796875,
+      "grad_norm": 0.862014651298523,
+      "learning_rate": 9.96259911996461e-05,
+      "loss": 0.03160874918103218,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.03211,
+      "step": 46,
+      "tokens/total": 1394688,
+      "tokens/train_per_sec_per_gpu": 34.54,
+      "tokens/trainable": 20967
+    },
+    {
+      "epoch": 0.18359375,
+      "grad_norm": 0.3395104706287384,
+      "learning_rate": 9.958771393851491e-05,
+      "loss": 0.0041851503774523735,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.98,
+      "memory/max_allocated (GiB)": 33.98,
+      "ppl": 1.00419,
+      "step": 47,
+      "tokens/total": 1425344,
+      "tokens/train_per_sec_per_gpu": 35.02,
+      "tokens/trainable": 21431
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 0.5284411311149597,
+      "learning_rate": 9.954758120060702e-05,
+      "loss": 0.008371120318770409,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00841,
+      "step": 48,
+      "tokens/total": 1455712,
+      "tokens/train_per_sec_per_gpu": 35.22,
+      "tokens/trainable": 21854
+    },
+    {
+      "epoch": 0.19140625,
+      "grad_norm": 7.4576287269592285,
+      "learning_rate": 9.950559465600948e-05,
+      "loss": 0.06921376287937164,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.29,
+      "memory/max_allocated (GiB)": 33.29,
+      "ppl": 1.07167,
+      "step": 49,
+      "tokens/total": 1483760,
+      "tokens/train_per_sec_per_gpu": 32.07,
+      "tokens/trainable": 22269
+    },
+    {
+      "epoch": 0.1953125,
+      "grad_norm": 0.8116379976272583,
+      "learning_rate": 9.946175605195379e-05,
+      "loss": 0.018268652260303497,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01844,
+      "step": 50,
+      "tokens/total": 1514128,
+      "tokens/train_per_sec_per_gpu": 36.71,
+      "tokens/trainable": 22716
+    },
+    {
+      "epoch": 0.19921875,
+      "grad_norm": 0.6055048108100891,
+      "learning_rate": 9.941606721274322e-05,
+      "loss": 0.01631464995443821,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01645,
+      "step": 51,
+      "tokens/total": 1544368,
+      "tokens/train_per_sec_per_gpu": 32.46,
+      "tokens/trainable": 23164
+    },
+    {
+      "epoch": 0.203125,
+      "grad_norm": 3.1200973987579346,
+      "learning_rate": 9.936853003967685e-05,
+      "loss": 0.021983586251735687,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.02223,
+      "step": 52,
+      "tokens/total": 1574672,
+      "tokens/train_per_sec_per_gpu": 37.43,
+      "tokens/trainable": 23659
+    },
+    {
+      "epoch": 0.20703125,
+      "grad_norm": 0.8921111822128296,
+      "learning_rate": 9.93191465109705e-05,
+      "loss": 0.028822191059589386,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02924,
+      "step": 53,
+      "tokens/total": 1604912,
+      "tokens/train_per_sec_per_gpu": 34.64,
+      "tokens/trainable": 24125
+    },
+    {
+      "epoch": 0.2109375,
+      "grad_norm": 0.619780957698822,
+      "learning_rate": 9.926791868167438e-05,
+      "loss": 0.01339790876954794,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.66,
+      "memory/max_allocated (GiB)": 33.66,
+      "ppl": 1.01349,
+      "step": 54,
+      "tokens/total": 1634784,
+      "tokens/train_per_sec_per_gpu": 38.31,
+      "tokens/trainable": 24599
+    },
+    {
+      "epoch": 0.21484375,
+      "grad_norm": 0.6819702982902527,
+      "learning_rate": 9.921484868358753e-05,
+      "loss": 0.023619044572114944,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.0239,
+      "step": 55,
+      "tokens/total": 1665280,
+      "tokens/train_per_sec_per_gpu": 34.33,
+      "tokens/trainable": 25096
+    },
+    {
+      "epoch": 0.21875,
+      "grad_norm": 0.11644987016916275,
+      "learning_rate": 9.915993872516924e-05,
+      "loss": 0.002859455067664385,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00286,
+      "step": 56,
+      "tokens/total": 1695616,
+      "tokens/train_per_sec_per_gpu": 36.96,
+      "tokens/trainable": 25570
+    },
+    {
+      "epoch": 0.22265625,
+      "grad_norm": 0.17120730876922607,
+      "learning_rate": 9.9103191091447e-05,
+      "loss": 0.0029868781566619873,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00299,
+      "step": 57,
+      "tokens/total": 1725824,
+      "tokens/train_per_sec_per_gpu": 33.57,
+      "tokens/trainable": 25994
+    },
+    {
+      "epoch": 0.2265625,
+      "grad_norm": 0.48202383518218994,
+      "learning_rate": 9.904460814392147e-05,
+      "loss": 0.012116724625229836,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.01219,
+      "step": 58,
+      "tokens/total": 1756336,
+      "tokens/train_per_sec_per_gpu": 34.27,
+      "tokens/trainable": 26438
+    },
+    {
+      "epoch": 0.23046875,
+      "grad_norm": 0.5425836443901062,
+      "learning_rate": 9.898419232046825e-05,
+      "loss": 0.012017138302326202,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01209,
+      "step": 59,
+      "tokens/total": 1786960,
+      "tokens/train_per_sec_per_gpu": 34.46,
+      "tokens/trainable": 26912
+    },
+    {
+      "epoch": 0.234375,
+      "grad_norm": 0.31607577204704285,
+      "learning_rate": 9.892194613523633e-05,
+      "loss": 0.004907170310616493,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00492,
+      "step": 60,
+      "tokens/total": 1817200,
+      "tokens/train_per_sec_per_gpu": 34.13,
+      "tokens/trainable": 27352
+    },
+    {
+      "epoch": 0.23828125,
+      "grad_norm": 0.2111223191022873,
+      "learning_rate": 9.885787217854357e-05,
+      "loss": 0.002509043551981449,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00251,
+      "step": 61,
+      "tokens/total": 1847808,
+      "tokens/train_per_sec_per_gpu": 39.56,
+      "tokens/trainable": 27859
+    },
+    {
+      "epoch": 0.2421875,
+      "grad_norm": 1.0228017568588257,
+      "learning_rate": 9.879197311676887e-05,
+      "loss": 0.015141930431127548,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 34.02,
+      "memory/max_allocated (GiB)": 34.02,
+      "ppl": 1.01526,
+      "step": 62,
+      "tokens/total": 1878416,
+      "tokens/train_per_sec_per_gpu": 32.31,
+      "tokens/trainable": 28310
+    },
+    {
+      "epoch": 0.24609375,
+      "grad_norm": 0.30590832233428955,
+      "learning_rate": 9.872425169224113e-05,
+      "loss": 0.0028732307255268097,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00288,
+      "step": 63,
+      "tokens/total": 1908896,
+      "tokens/train_per_sec_per_gpu": 32.61,
+      "tokens/trainable": 28772
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.7455278635025024,
+      "learning_rate": 9.865471072312528e-05,
+      "loss": 0.02393944375216961,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02423,
+      "step": 64,
+      "tokens/total": 1939200,
+      "tokens/train_per_sec_per_gpu": 33.63,
+      "tokens/trainable": 29220
+    },
+    {
+      "epoch": 0.25390625,
+      "grad_norm": 0.23692572116851807,
+      "learning_rate": 9.858335310330492e-05,
+      "loss": 0.0020673489198088646,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00207,
+      "step": 65,
+      "tokens/total": 1969696,
+      "tokens/train_per_sec_per_gpu": 31.87,
+      "tokens/trainable": 29685
+    },
+    {
+      "epoch": 0.2578125,
+      "grad_norm": 0.2896324694156647,
+      "learning_rate": 9.851018180226185e-05,
+      "loss": 0.013750756159424782,
+      "memory/device_reserved (GiB)": 35.06,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01385,
+      "step": 66,
+      "tokens/total": 2000064,
+      "tokens/train_per_sec_per_gpu": 33.79,
+      "tokens/trainable": 30112
+    },
+    {
+      "epoch": 0.26171875,
+      "grad_norm": 0.44811347126960754,
+      "learning_rate": 9.843519986495259e-05,
+      "loss": 0.03262199088931084,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.03316,
+      "step": 67,
+      "tokens/total": 2028288,
+      "tokens/train_per_sec_per_gpu": 37.96,
+      "tokens/trainable": 30591
+    },
+    {
+      "epoch": 0.265625,
+      "grad_norm": 0.5987648367881775,
+      "learning_rate": 9.835841041168162e-05,
+      "loss": 0.006292116362601519,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00631,
+      "step": 68,
+      "tokens/total": 2058736,
+      "tokens/train_per_sec_per_gpu": 35.47,
+      "tokens/trainable": 31070
+    },
+    {
+      "epoch": 0.26953125,
+      "grad_norm": 0.3963381052017212,
+      "learning_rate": 9.82798166379715e-05,
+      "loss": 0.004650093615055084,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00466,
+      "step": 69,
+      "tokens/total": 2089040,
+      "tokens/train_per_sec_per_gpu": 34.27,
+      "tokens/trainable": 31550
+    },
+    {
+      "epoch": 0.2734375,
+      "grad_norm": 1.0574487447738647,
+      "learning_rate": 9.819942181443002e-05,
+      "loss": 0.019242461770772934,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01943,
+      "step": 70,
+      "tokens/total": 2119200,
+      "tokens/train_per_sec_per_gpu": 32.63,
+      "tokens/trainable": 32001
+    },
+    {
+      "epoch": 0.27734375,
+      "grad_norm": 0.44274845719337463,
+      "learning_rate": 9.811722928661392e-05,
+      "loss": 0.007928689941763878,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.35,
+      "memory/max_allocated (GiB)": 33.35,
+      "ppl": 1.00796,
+      "step": 71,
+      "tokens/total": 2147440,
+      "tokens/train_per_sec_per_gpu": 32.29,
+      "tokens/trainable": 32417
+    },
+    {
+      "epoch": 0.28125,
+      "grad_norm": 0.4176182150840759,
+      "learning_rate": 9.803324247488975e-05,
+      "loss": 0.007203593384474516,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00723,
+      "step": 72,
+      "tokens/total": 2177792,
+      "tokens/train_per_sec_per_gpu": 34.82,
+      "tokens/trainable": 32876
+    },
+    {
+      "epoch": 0.28515625,
+      "grad_norm": 0.5054254531860352,
+      "learning_rate": 9.794746487429161e-05,
+      "loss": 0.014437871053814888,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01454,
+      "step": 73,
+      "tokens/total": 2208384,
+      "tokens/train_per_sec_per_gpu": 36.05,
+      "tokens/trainable": 33347
+    },
+    {
+      "epoch": 0.2890625,
+      "grad_norm": 0.57452392578125,
+      "learning_rate": 9.785990005437554e-05,
+      "loss": 0.010695832781493664,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01075,
+      "step": 74,
+      "tokens/total": 2238912,
+      "tokens/train_per_sec_per_gpu": 32.75,
+      "tokens/trainable": 33814
+    },
+    {
+      "epoch": 0.29296875,
+      "grad_norm": 1.2015029191970825,
+      "learning_rate": 9.777055165907117e-05,
+      "loss": 0.026340341195464134,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.02669,
+      "step": 75,
+      "tokens/total": 2269008,
+      "tokens/train_per_sec_per_gpu": 34.81,
+      "tokens/trainable": 34228
+    },
+    {
+      "epoch": 0.296875,
+      "grad_norm": 0.6986287236213684,
+      "learning_rate": 9.767942340652993e-05,
+      "loss": 0.00647423230111599,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.0065,
+      "step": 76,
+      "tokens/total": 2299264,
+      "tokens/train_per_sec_per_gpu": 36.41,
+      "tokens/trainable": 34717
+    },
+    {
+      "epoch": 0.30078125,
+      "grad_norm": 0.8190402984619141,
+      "learning_rate": 9.758651908897035e-05,
+      "loss": 0.04209361970424652,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.76,
+      "memory/max_allocated (GiB)": 33.76,
+      "ppl": 1.04299,
+      "step": 77,
+      "tokens/total": 2329360,
+      "tokens/train_per_sec_per_gpu": 32.06,
+      "tokens/trainable": 35144
+    },
+    {
+      "epoch": 0.3046875,
+      "grad_norm": 0.31275519728660583,
+      "learning_rate": 9.749184257252033e-05,
+      "loss": 0.016447898000478745,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.01658,
+      "step": 78,
+      "tokens/total": 2359872,
+      "tokens/train_per_sec_per_gpu": 35.18,
+      "tokens/trainable": 35608
+    },
+    {
+      "epoch": 0.30859375,
+      "grad_norm": 0.5485185384750366,
+      "learning_rate": 9.739539779705614e-05,
+      "loss": 0.027634259313344955,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02802,
+      "step": 79,
+      "tokens/total": 2390224,
+      "tokens/train_per_sec_per_gpu": 29.79,
+      "tokens/trainable": 36035
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 0.3298993408679962,
+      "learning_rate": 9.729718877603861e-05,
+      "loss": 0.01857740990817547,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01875,
+      "step": 80,
+      "tokens/total": 2420720,
+      "tokens/train_per_sec_per_gpu": 38.66,
+      "tokens/trainable": 36520
+    },
+    {
+      "epoch": 0.31640625,
+      "grad_norm": 0.19593212008476257,
+      "learning_rate": 9.719721959634592e-05,
+      "loss": 0.003864901140332222,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00387,
+      "step": 81,
+      "tokens/total": 2451040,
+      "tokens/train_per_sec_per_gpu": 33.51,
+      "tokens/trainable": 36977
+    },
+    {
+      "epoch": 0.3203125,
+      "grad_norm": 0.26418155431747437,
+      "learning_rate": 9.709549441810375e-05,
+      "loss": 0.014170816168189049,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.44,
+      "memory/max_allocated (GiB)": 33.44,
+      "ppl": 1.01427,
+      "step": 82,
+      "tokens/total": 2479504,
+      "tokens/train_per_sec_per_gpu": 33.68,
+      "tokens/trainable": 37437
+    },
+    {
+      "epoch": 0.32421875,
+      "grad_norm": 0.4824850261211395,
+      "learning_rate": 9.699201747451195e-05,
+      "loss": 0.01662585139274597,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01676,
+      "step": 83,
+      "tokens/total": 2509776,
+      "tokens/train_per_sec_per_gpu": 33.38,
+      "tokens/trainable": 37889
+    },
+    {
+      "epoch": 0.328125,
+      "grad_norm": 0.7037883400917053,
+      "learning_rate": 9.688679307166854e-05,
+      "loss": 0.03413955122232437,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.03473,
+      "step": 84,
+      "tokens/total": 2540272,
+      "tokens/train_per_sec_per_gpu": 33.15,
+      "tokens/trainable": 38352
+    },
+    {
+      "epoch": 0.33203125,
+      "grad_norm": 0.18567293882369995,
+      "learning_rate": 9.677982558839042e-05,
+      "loss": 0.005708993412554264,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.69,
+      "memory/max_allocated (GiB)": 33.69,
+      "ppl": 1.00573,
+      "step": 85,
+      "tokens/total": 2570512,
+      "tokens/train_per_sec_per_gpu": 34.01,
+      "tokens/trainable": 38807
+    },
+    {
+      "epoch": 0.3359375,
+      "grad_norm": 0.41593775153160095,
+      "learning_rate": 9.66711194760312e-05,
+      "loss": 0.02402549795806408,
+      "memory/device_reserved (GiB)": 35.92,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.02432,
+      "step": 86,
+      "tokens/total": 2600992,
+      "tokens/train_per_sec_per_gpu": 30.82,
+      "tokens/trainable": 39235
+    },
+    {
+      "epoch": 0.33984375,
+      "grad_norm": 0.851169228553772,
+      "learning_rate": 9.656067925829593e-05,
+      "loss": 0.05195997655391693,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.05333,
+      "step": 87,
+      "tokens/total": 2631360,
+      "tokens/train_per_sec_per_gpu": 36.86,
+      "tokens/trainable": 39743
+    },
+    {
+      "epoch": 0.34375,
+      "grad_norm": 0.29017335176467896,
+      "learning_rate": 9.644850953105288e-05,
+      "loss": 0.01987021416425705,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.02007,
+      "step": 88,
+      "tokens/total": 2661520,
+      "tokens/train_per_sec_per_gpu": 33.25,
+      "tokens/trainable": 40183
+    },
+    {
+      "epoch": 0.34765625,
+      "grad_norm": 0.8229841589927673,
+      "learning_rate": 9.633461496214225e-05,
+      "loss": 0.026500295847654343,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02685,
+      "step": 89,
+      "tokens/total": 2691984,
+      "tokens/train_per_sec_per_gpu": 33.22,
+      "tokens/trainable": 40627
+    },
+    {
+      "epoch": 0.3515625,
+      "grad_norm": 0.681232213973999,
+      "learning_rate": 9.621900029118195e-05,
+      "loss": 0.013729426078498363,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01382,
+      "step": 90,
+      "tokens/total": 2722272,
+      "tokens/train_per_sec_per_gpu": 31.25,
+      "tokens/trainable": 41045
+    },
+    {
+      "epoch": 0.35546875,
+      "grad_norm": 0.29015299677848816,
+      "learning_rate": 9.610167032937036e-05,
+      "loss": 0.008784668520092964,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00882,
+      "step": 91,
+      "tokens/total": 2752528,
+      "tokens/train_per_sec_per_gpu": 36.64,
+      "tokens/trainable": 41510
+    },
+    {
+      "epoch": 0.359375,
+      "grad_norm": 0.24605843424797058,
+      "learning_rate": 9.598262995928611e-05,
+      "loss": 0.013710295781493187,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.0138,
+      "step": 92,
+      "tokens/total": 2782960,
+      "tokens/train_per_sec_per_gpu": 35.99,
+      "tokens/trainable": 41967
+    },
+    {
+      "epoch": 0.36328125,
+      "grad_norm": 1.061617374420166,
+      "learning_rate": 9.586188413468492e-05,
+      "loss": 0.028332654386758804,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02874,
+      "step": 93,
+      "tokens/total": 2813408,
+      "tokens/train_per_sec_per_gpu": 35.22,
+      "tokens/trainable": 42437
+    },
+    {
+      "epoch": 0.3671875,
+      "grad_norm": 0.6703851819038391,
+      "learning_rate": 9.57394378802934e-05,
+      "loss": 0.010767554864287376,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01083,
+      "step": 94,
+      "tokens/total": 2843888,
+      "tokens/train_per_sec_per_gpu": 36.99,
+      "tokens/trainable": 42923
+    },
+    {
+      "epoch": 0.37109375,
+      "grad_norm": 0.6564542651176453,
+      "learning_rate": 9.56152962916e-05,
+      "loss": 0.004970403853803873,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00498,
+      "step": 95,
+      "tokens/total": 2874240,
+      "tokens/train_per_sec_per_gpu": 39.09,
+      "tokens/trainable": 43421
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 0.5388283729553223,
+      "learning_rate": 9.548946453464296e-05,
+      "loss": 0.013605385087430477,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.0137,
+      "step": 96,
+      "tokens/total": 2904576,
+      "tokens/train_per_sec_per_gpu": 33.33,
+      "tokens/trainable": 43864
+    },
+    {
+      "epoch": 0.37890625,
+      "grad_norm": 0.05957590416073799,
+      "learning_rate": 9.53619478457953e-05,
+      "loss": 0.0009635845199227333,
+      "memory/device_reserved (GiB)": 35.94,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00096,
+      "step": 97,
+      "tokens/total": 2935040,
+      "tokens/train_per_sec_per_gpu": 31.85,
+      "tokens/trainable": 44295
+    },
+    {
+      "epoch": 0.3828125,
+      "grad_norm": 0.6069821715354919,
+      "learning_rate": 9.523275153154695e-05,
+      "loss": 0.018155831843614578,
+      "memory/device_reserved (GiB)": 35.19,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01832,
+      "step": 98,
+      "tokens/total": 2965360,
+      "tokens/train_per_sec_per_gpu": 39.17,
+      "tokens/trainable": 44786
+    },
+    {
+      "epoch": 0.38671875,
+      "grad_norm": 0.8626548647880554,
+      "learning_rate": 9.51018809682839e-05,
+      "loss": 0.010867835953831673,
+      "memory/device_reserved (GiB)": 35.76,
+      "memory/max_active (GiB)": 34.03,
+      "memory/max_allocated (GiB)": 34.03,
+      "ppl": 1.01093,
+      "step": 99,
+      "tokens/total": 2995952,
+      "tokens/train_per_sec_per_gpu": 36.55,
+      "tokens/trainable": 45249
+    },
+    {
+      "epoch": 0.390625,
+      "grad_norm": 0.25867366790771484,
+      "learning_rate": 9.49693416020645e-05,
+      "loss": 0.0033954819664359093,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.0034,
+      "step": 100,
+      "tokens/total": 3026320,
+      "tokens/train_per_sec_per_gpu": 33.38,
+      "tokens/trainable": 45693
+    },
+    {
+      "epoch": 0.39453125,
+      "grad_norm": 0.6113082766532898,
+      "learning_rate": 9.483513894839276e-05,
+      "loss": 0.0068494766019284725,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00687,
+      "step": 101,
+      "tokens/total": 3056560,
+      "tokens/train_per_sec_per_gpu": 33.34,
+      "tokens/trainable": 46136
+    },
+    {
+      "epoch": 0.3984375,
+      "grad_norm": 1.2137420177459717,
+      "learning_rate": 9.469927859198888e-05,
+      "loss": 0.01495033223181963,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01506,
+      "step": 102,
+      "tokens/total": 3087136,
+      "tokens/train_per_sec_per_gpu": 32.91,
+      "tokens/trainable": 46592
+    },
+    {
+      "epoch": 0.40234375,
+      "grad_norm": 0.34507185220718384,
+      "learning_rate": 9.456176618655689e-05,
+      "loss": 0.006014987826347351,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00603,
+      "step": 103,
+      "tokens/total": 3117712,
+      "tokens/train_per_sec_per_gpu": 32.3,
+      "tokens/trainable": 47059
+    },
+    {
+      "epoch": 0.40625,
+      "grad_norm": 1.0635204315185547,
+      "learning_rate": 9.442260745454927e-05,
+      "loss": 0.015764378011226654,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01589,
+      "step": 104,
+      "tokens/total": 3147984,
+      "tokens/train_per_sec_per_gpu": 38.13,
+      "tokens/trainable": 47548
+    },
+    {
+      "epoch": 0.41015625,
+      "grad_norm": 0.06298446655273438,
+      "learning_rate": 9.428180818692884e-05,
+      "loss": 0.0005863154074177146,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00059,
+      "step": 105,
+      "tokens/total": 3178352,
+      "tokens/train_per_sec_per_gpu": 36.72,
+      "tokens/trainable": 48041
+    },
+    {
+      "epoch": 0.4140625,
+      "grad_norm": 0.2887956202030182,
+      "learning_rate": 9.413937424292791e-05,
+      "loss": 0.0026632004883140326,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.00267,
+      "step": 106,
+      "tokens/total": 3206608,
+      "tokens/train_per_sec_per_gpu": 34.48,
+      "tokens/trainable": 48492
+    },
+    {
+      "epoch": 0.41796875,
+      "grad_norm": 1.3830534219741821,
+      "learning_rate": 9.399531154980424e-05,
+      "loss": 0.05919238552451134,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.06098,
+      "step": 107,
+      "tokens/total": 3236704,
+      "tokens/train_per_sec_per_gpu": 31.08,
+      "tokens/trainable": 48926
+    },
+    {
+      "epoch": 0.421875,
+      "grad_norm": 0.05409906059503555,
+      "learning_rate": 9.384962610259455e-05,
+      "loss": 0.0008346759132109582,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00084,
+      "step": 108,
+      "tokens/total": 3267072,
+      "tokens/train_per_sec_per_gpu": 38.28,
+      "tokens/trainable": 49397
+    },
+    {
+      "epoch": 0.42578125,
+      "grad_norm": 0.4702659845352173,
+      "learning_rate": 9.370232396386494e-05,
+      "loss": 0.011177235282957554,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.01124,
+      "step": 109,
+      "tokens/total": 3297200,
+      "tokens/train_per_sec_per_gpu": 37.12,
+      "tokens/trainable": 49864
+    },
+    {
+      "epoch": 0.4296875,
+      "grad_norm": 0.36087653040885925,
+      "learning_rate": 9.355341126345868e-05,
+      "loss": 0.0047053806483745575,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.00472,
+      "step": 110,
+      "tokens/total": 3327616,
+      "tokens/train_per_sec_per_gpu": 33.15,
+      "tokens/trainable": 50303
+    },
+    {
+      "epoch": 0.43359375,
+      "grad_norm": 0.4973398745059967,
+      "learning_rate": 9.340289419824107e-05,
+      "loss": 0.02818290702998638,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.02858,
+      "step": 111,
+      "tokens/total": 3357824,
+      "tokens/train_per_sec_per_gpu": 33.77,
+      "tokens/trainable": 50746
+    },
+    {
+      "epoch": 0.4375,
+      "grad_norm": 0.27993273735046387,
+      "learning_rate": 9.325077903184159e-05,
+      "loss": 0.007751852739602327,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.00778,
+      "step": 112,
+      "tokens/total": 3387856,
+      "tokens/train_per_sec_per_gpu": 31.99,
+      "tokens/trainable": 51192
+    },
+    {
+      "epoch": 0.44140625,
+      "grad_norm": 1.2636487483978271,
+      "learning_rate": 9.30970720943932e-05,
+      "loss": 0.006412792485207319,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00643,
+      "step": 113,
+      "tokens/total": 3418304,
+      "tokens/train_per_sec_per_gpu": 33.1,
+      "tokens/trainable": 51648
+    },
+    {
+      "epoch": 0.4453125,
+      "grad_norm": 0.34893789887428284,
+      "learning_rate": 9.2941779782269e-05,
+      "loss": 0.010952373966574669,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.01101,
+      "step": 114,
+      "tokens/total": 3448560,
+      "tokens/train_per_sec_per_gpu": 30.78,
+      "tokens/trainable": 52061
+    },
+    {
+      "epoch": 0.44921875,
+      "grad_norm": 0.2650280296802521,
+      "learning_rate": 9.278490855781596e-05,
+      "loss": 0.007384698837995529,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00741,
+      "step": 115,
+      "tokens/total": 3478928,
+      "tokens/train_per_sec_per_gpu": 36.65,
+      "tokens/trainable": 52550
+    },
+    {
+      "epoch": 0.453125,
+      "grad_norm": 0.2596147656440735,
+      "learning_rate": 9.262646494908604e-05,
+      "loss": 0.004922072868794203,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00493,
+      "step": 116,
+      "tokens/total": 3509232,
+      "tokens/train_per_sec_per_gpu": 38.88,
+      "tokens/trainable": 53051
+    },
+    {
+      "epoch": 0.45703125,
+      "grad_norm": 0.24072641134262085,
+      "learning_rate": 9.246645554956457e-05,
+      "loss": 0.003113753627985716,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.00312,
+      "step": 117,
+      "tokens/total": 3539680,
+      "tokens/train_per_sec_per_gpu": 35.76,
+      "tokens/trainable": 53510
+    },
+    {
+      "epoch": 0.4609375,
+      "grad_norm": 0.27470293641090393,
+      "learning_rate": 9.230488701789578e-05,
+      "loss": 0.010554850101470947,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01061,
+      "step": 118,
+      "tokens/total": 3570320,
+      "tokens/train_per_sec_per_gpu": 33.21,
+      "tokens/trainable": 53966
+    },
+    {
+      "epoch": 0.46484375,
+      "grad_norm": 0.6720938682556152,
+      "learning_rate": 9.214176607760577e-05,
+      "loss": 0.021780190989375114,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.02202,
+      "step": 119,
+      "tokens/total": 3600384,
+      "tokens/train_per_sec_per_gpu": 34.72,
+      "tokens/trainable": 54399
+    },
+    {
+      "epoch": 0.46875,
+      "grad_norm": 0.5802715420722961,
+      "learning_rate": 9.197709951682268e-05,
+      "loss": 0.022010700777173042,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02225,
+      "step": 120,
+      "tokens/total": 3630672,
+      "tokens/train_per_sec_per_gpu": 36.96,
+      "tokens/trainable": 54875
+    },
+    {
+      "epoch": 0.47265625,
+      "grad_norm": 0.2093982845544815,
+      "learning_rate": 9.181089418799428e-05,
+      "loss": 0.0025330549106001854,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00254,
+      "step": 121,
+      "tokens/total": 3660992,
+      "tokens/train_per_sec_per_gpu": 33.01,
+      "tokens/trainable": 55294
+    },
+    {
+      "epoch": 0.4765625,
+      "grad_norm": 0.22755542397499084,
+      "learning_rate": 9.164315700760271e-05,
+      "loss": 0.008100905455648899,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.00813,
+      "step": 122,
+      "tokens/total": 3691248,
+      "tokens/train_per_sec_per_gpu": 31.93,
+      "tokens/trainable": 55734
+    },
+    {
+      "epoch": 0.48046875,
+      "grad_norm": 0.5118339657783508,
+      "learning_rate": 9.147389495587671e-05,
+      "loss": 0.012025108560919762,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.0121,
+      "step": 123,
+      "tokens/total": 3721248,
+      "tokens/train_per_sec_per_gpu": 33.8,
+      "tokens/trainable": 56178
+    },
+    {
+      "epoch": 0.484375,
+      "grad_norm": 0.4289400279521942,
+      "learning_rate": 9.130311507650116e-05,
+      "loss": 0.01151777058839798,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01158,
+      "step": 124,
+      "tokens/total": 3751440,
+      "tokens/train_per_sec_per_gpu": 33.49,
+      "tokens/trainable": 56623
+    },
+    {
+      "epoch": 0.48828125,
+      "grad_norm": 0.12747441232204437,
+      "learning_rate": 9.113082447632394e-05,
+      "loss": 0.0026185065507888794,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00262,
+      "step": 125,
+      "tokens/total": 3781696,
+      "tokens/train_per_sec_per_gpu": 37.07,
+      "tokens/trainable": 57115
+    },
+    {
+      "epoch": 0.4921875,
+      "grad_norm": 0.32073503732681274,
+      "learning_rate": 9.09570303250602e-05,
+      "loss": 0.004942174069583416,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00495,
+      "step": 126,
+      "tokens/total": 3812096,
+      "tokens/train_per_sec_per_gpu": 33.23,
+      "tokens/trainable": 57561
+    },
+    {
+      "epoch": 0.49609375,
+      "grad_norm": 0.764901340007782,
+      "learning_rate": 9.078173985499394e-05,
+      "loss": 0.023843681439757347,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02413,
+      "step": 127,
+      "tokens/total": 3842720,
+      "tokens/train_per_sec_per_gpu": 35.17,
+      "tokens/trainable": 58039
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.43675562739372253,
+      "learning_rate": 9.060496036067713e-05,
+      "loss": 0.012037287466228008,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.01211,
+      "step": 128,
+      "tokens/total": 3871072,
+      "tokens/train_per_sec_per_gpu": 35.55,
+      "tokens/trainable": 58488
+    },
+    {
+      "epoch": 0.50390625,
+      "grad_norm": 0.36475399136543274,
+      "learning_rate": 9.042669919862615e-05,
+      "loss": 0.007448950316756964,
+      "memory/device_reserved (GiB)": 36.01,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.00748,
+      "step": 129,
+      "tokens/total": 3901504,
+      "tokens/train_per_sec_per_gpu": 33.0,
+      "tokens/trainable": 58938
+    },
+    {
+      "epoch": 0.5078125,
+      "grad_norm": 0.2994728982448578,
+      "learning_rate": 9.024696378701557e-05,
+      "loss": 0.005909992381930351,
+      "memory/device_reserved (GiB)": 36.01,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00593,
+      "step": 130,
+      "tokens/total": 3931968,
+      "tokens/train_per_sec_per_gpu": 37.36,
+      "tokens/trainable": 59406
+    },
+    {
+      "epoch": 0.51171875,
+      "grad_norm": 0.41924577951431274,
+      "learning_rate": 9.006576160536948e-05,
+      "loss": 0.015188692137598991,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.0153,
+      "step": 131,
+      "tokens/total": 3962048,
+      "tokens/train_per_sec_per_gpu": 32.61,
+      "tokens/trainable": 59842
+    },
+    {
+      "epoch": 0.515625,
+      "grad_norm": 0.17138168215751648,
+      "learning_rate": 8.988310019425035e-05,
+      "loss": 0.003963212016969919,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00397,
+      "step": 132,
+      "tokens/total": 3992512,
+      "tokens/train_per_sec_per_gpu": 36.14,
+      "tokens/trainable": 60316
+    },
+    {
+      "epoch": 0.51953125,
+      "grad_norm": 0.6236148476600647,
+      "learning_rate": 8.969898715494506e-05,
+      "loss": 0.01790568232536316,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01807,
+      "step": 133,
+      "tokens/total": 4022864,
+      "tokens/train_per_sec_per_gpu": 35.89,
+      "tokens/trainable": 60809
+    },
+    {
+      "epoch": 0.5234375,
+      "grad_norm": 0.41700077056884766,
+      "learning_rate": 8.951343014914869e-05,
+      "loss": 0.004474501125514507,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.00448,
+      "step": 134,
+      "tokens/total": 4051344,
+      "tokens/train_per_sec_per_gpu": 32.08,
+      "tokens/trainable": 61254
+    },
+    {
+      "epoch": 0.52734375,
+      "grad_norm": 0.6247485280036926,
+      "learning_rate": 8.932643689864568e-05,
+      "loss": 0.011853284202516079,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01192,
+      "step": 135,
+      "tokens/total": 4081728,
+      "tokens/train_per_sec_per_gpu": 36.49,
+      "tokens/trainable": 61725
+    },
+    {
+      "epoch": 0.53125,
+      "grad_norm": 0.3736969530582428,
+      "learning_rate": 8.913801518498845e-05,
+      "loss": 0.0037904747296124697,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.7,
+      "memory/max_allocated (GiB)": 33.7,
+      "ppl": 1.0038,
+      "step": 136,
+      "tokens/total": 4111904,
+      "tokens/train_per_sec_per_gpu": 38.69,
+      "tokens/trainable": 62200
+    },
+    {
+      "epoch": 0.53515625,
+      "grad_norm": 0.15076029300689697,
+      "learning_rate": 8.894817284917364e-05,
+      "loss": 0.0031103340443223715,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00312,
+      "step": 137,
+      "tokens/total": 4142208,
+      "tokens/train_per_sec_per_gpu": 36.91,
+      "tokens/trainable": 62698
+    },
+    {
+      "epoch": 0.5390625,
+      "grad_norm": 0.6855320334434509,
+      "learning_rate": 8.875691779131569e-05,
+      "loss": 0.027959343045949936,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.02835,
+      "step": 138,
+      "tokens/total": 4172512,
+      "tokens/train_per_sec_per_gpu": 37.03,
+      "tokens/trainable": 63158
+    },
+    {
+      "epoch": 0.54296875,
+      "grad_norm": 0.42463886737823486,
+      "learning_rate": 8.856425797031829e-05,
+      "loss": 0.008795550093054771,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00883,
+      "step": 139,
+      "tokens/total": 4202944,
+      "tokens/train_per_sec_per_gpu": 36.87,
+      "tokens/trainable": 63638
+    },
+    {
+      "epoch": 0.546875,
+      "grad_norm": 0.7433419227600098,
+      "learning_rate": 8.837020140354295e-05,
+      "loss": 0.017757249996066093,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.42,
+      "memory/max_allocated (GiB)": 33.42,
+      "ppl": 1.01792,
+      "step": 140,
+      "tokens/total": 4231264,
+      "tokens/train_per_sec_per_gpu": 33.47,
+      "tokens/trainable": 64055
+    },
+    {
+      "epoch": 0.55078125,
+      "grad_norm": 0.16250354051589966,
+      "learning_rate": 8.817475616647554e-05,
+      "loss": 0.0028627081774175167,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00287,
+      "step": 141,
+      "tokens/total": 4261488,
+      "tokens/train_per_sec_per_gpu": 35.54,
+      "tokens/trainable": 64512
+    },
+    {
+      "epoch": 0.5546875,
+      "grad_norm": 0.7241039872169495,
+      "learning_rate": 8.797793039239017e-05,
+      "loss": 0.0090868528932333,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.00913,
+      "step": 142,
+      "tokens/total": 4292032,
+      "tokens/train_per_sec_per_gpu": 34.22,
+      "tokens/trainable": 64950
+    },
+    {
+      "epoch": 0.55859375,
+      "grad_norm": 0.26554936170578003,
+      "learning_rate": 8.777973227201069e-05,
+      "loss": 0.008465563878417015,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.0085,
+      "step": 143,
+      "tokens/total": 4322032,
+      "tokens/train_per_sec_per_gpu": 38.11,
+      "tokens/trainable": 65429
+    },
+    {
+      "epoch": 0.5625,
+      "grad_norm": 0.12835027277469635,
+      "learning_rate": 8.758017005316988e-05,
+      "loss": 0.0033488553017377853,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00335,
+      "step": 144,
+      "tokens/total": 4352128,
+      "tokens/train_per_sec_per_gpu": 37.15,
+      "tokens/trainable": 65895
+    },
+    {
+      "epoch": 0.56640625,
+      "grad_norm": 0.5516126751899719,
+      "learning_rate": 8.737925204046629e-05,
+      "loss": 0.01850893907248974,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.01868,
+      "step": 145,
+      "tokens/total": 4382368,
+      "tokens/train_per_sec_per_gpu": 32.01,
+      "tokens/trainable": 66327
+    },
+    {
+      "epoch": 0.5703125,
+      "grad_norm": 0.3258255124092102,
+      "learning_rate": 8.717698659491851e-05,
+      "loss": 0.008647755719721317,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.00869,
+      "step": 146,
+      "tokens/total": 4412640,
+      "tokens/train_per_sec_per_gpu": 31.94,
+      "tokens/trainable": 66778
+    },
+    {
+      "epoch": 0.57421875,
+      "grad_norm": 0.3512382209300995,
+      "learning_rate": 8.697338213361735e-05,
+      "loss": 0.010574431158602238,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01063,
+      "step": 147,
+      "tokens/total": 4442896,
+      "tokens/train_per_sec_per_gpu": 35.74,
+      "tokens/trainable": 67192
+    },
+    {
+      "epoch": 0.578125,
+      "grad_norm": 0.2663547396659851,
+      "learning_rate": 8.676844712937552e-05,
+      "loss": 0.0057985298335552216,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00582,
+      "step": 148,
+      "tokens/total": 4473248,
+      "tokens/train_per_sec_per_gpu": 35.81,
+      "tokens/trainable": 67677
+    },
+    {
+      "epoch": 0.58203125,
+      "grad_norm": 0.36537718772888184,
+      "learning_rate": 8.656219011037509e-05,
+      "loss": 0.007628970313817263,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00766,
+      "step": 149,
+      "tokens/total": 4503664,
+      "tokens/train_per_sec_per_gpu": 34.29,
+      "tokens/trainable": 68121
+    },
+    {
+      "epoch": 0.5859375,
+      "grad_norm": 0.3554728031158447,
+      "learning_rate": 8.63546196598125e-05,
+      "loss": 0.008026087656617165,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00806,
+      "step": 150,
+      "tokens/total": 4533760,
+      "tokens/train_per_sec_per_gpu": 33.65,
+      "tokens/trainable": 68589
+    },
+    {
+      "epoch": 0.58984375,
+      "grad_norm": 0.34241166710853577,
+      "learning_rate": 8.614574441554145e-05,
+      "loss": 0.0072187017649412155,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00724,
+      "step": 151,
+      "tokens/total": 4563872,
+      "tokens/train_per_sec_per_gpu": 30.46,
+      "tokens/trainable": 69055
+    },
+    {
+      "epoch": 0.59375,
+      "grad_norm": 0.24926632642745972,
+      "learning_rate": 8.593557306971349e-05,
+      "loss": 0.00544877490028739,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.00546,
+      "step": 152,
+      "tokens/total": 4594384,
+      "tokens/train_per_sec_per_gpu": 35.61,
+      "tokens/trainable": 69501
+    },
+    {
+      "epoch": 0.59765625,
+      "grad_norm": 0.05485764890909195,
+      "learning_rate": 8.572411436841618e-05,
+      "loss": 0.0006452050292864442,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00065,
+      "step": 153,
+      "tokens/total": 4624816,
+      "tokens/train_per_sec_per_gpu": 34.3,
+      "tokens/trainable": 69964
+    },
+    {
+      "epoch": 0.6015625,
+      "grad_norm": 0.3729817569255829,
+      "learning_rate": 8.551137711130922e-05,
+      "loss": 0.0017574160592630506,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00176,
+      "step": 154,
+      "tokens/total": 4655136,
+      "tokens/train_per_sec_per_gpu": 37.11,
+      "tokens/trainable": 70431
+    },
+    {
+      "epoch": 0.60546875,
+      "grad_norm": 0.5586419701576233,
+      "learning_rate": 8.529737015125824e-05,
+      "loss": 0.02433975785970688,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02464,
+      "step": 155,
+      "tokens/total": 4683488,
+      "tokens/train_per_sec_per_gpu": 37.73,
+      "tokens/trainable": 70875
+    },
+    {
+      "epoch": 0.609375,
+      "grad_norm": 0.3311072289943695,
+      "learning_rate": 8.508210239396639e-05,
+      "loss": 0.005622293334454298,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00564,
+      "step": 156,
+      "tokens/total": 4714032,
+      "tokens/train_per_sec_per_gpu": 34.32,
+      "tokens/trainable": 71333
+    },
+    {
+      "epoch": 0.61328125,
+      "grad_norm": 0.16862744092941284,
+      "learning_rate": 8.486558279760375e-05,
+      "loss": 0.0018788446905091405,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00188,
+      "step": 157,
+      "tokens/total": 4744192,
+      "tokens/train_per_sec_per_gpu": 35.42,
+      "tokens/trainable": 71795
+    },
+    {
+      "epoch": 0.6171875,
+      "grad_norm": 0.66270512342453,
+      "learning_rate": 8.464782037243449e-05,
+      "loss": 0.013912302441895008,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01401,
+      "step": 158,
+      "tokens/total": 4774592,
+      "tokens/train_per_sec_per_gpu": 32.56,
+      "tokens/trainable": 72231
+    },
+    {
+      "epoch": 0.62109375,
+      "grad_norm": 0.33048003911972046,
+      "learning_rate": 8.442882418044202e-05,
+      "loss": 0.011282861232757568,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.01135,
+      "step": 159,
+      "tokens/total": 4804672,
+      "tokens/train_per_sec_per_gpu": 30.89,
+      "tokens/trainable": 72660
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 0.22905726730823517,
+      "learning_rate": 8.420860333495179e-05,
+      "loss": 0.004656655248254538,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00467,
+      "step": 160,
+      "tokens/total": 4835056,
+      "tokens/train_per_sec_per_gpu": 31.96,
+      "tokens/trainable": 73115
+    },
+    {
+      "epoch": 0.62890625,
+      "grad_norm": 0.603830873966217,
+      "learning_rate": 8.398716700025208e-05,
+      "loss": 0.022144438698887825,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.02239,
+      "step": 161,
+      "tokens/total": 4865296,
+      "tokens/train_per_sec_per_gpu": 31.21,
+      "tokens/trainable": 73536
+    },
+    {
+      "epoch": 0.6328125,
+      "grad_norm": 0.3103669583797455,
+      "learning_rate": 8.376452439121266e-05,
+      "loss": 0.008616160601377487,
+      "memory/device_reserved (GiB)": 35.81,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.00865,
+      "step": 162,
+      "tokens/total": 4895872,
+      "tokens/train_per_sec_per_gpu": 32.97,
+      "tokens/trainable": 73985
+    },
+    {
+      "epoch": 0.63671875,
+      "grad_norm": 0.19615702331066132,
+      "learning_rate": 8.354068477290124e-05,
+      "loss": 0.01020321249961853,
+      "memory/device_reserved (GiB)": 35.81,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01026,
+      "step": 163,
+      "tokens/total": 4926192,
+      "tokens/train_per_sec_per_gpu": 35.74,
+      "tokens/trainable": 74449
+    },
+    {
+      "epoch": 0.640625,
+      "grad_norm": 0.133578360080719,
+      "learning_rate": 8.331565746019807e-05,
+      "loss": 0.008446885272860527,
+      "memory/device_reserved (GiB)": 35.81,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00848,
+      "step": 164,
+      "tokens/total": 4956784,
+      "tokens/train_per_sec_per_gpu": 33.76,
+      "tokens/trainable": 74900
+    },
+    {
+      "epoch": 0.64453125,
+      "grad_norm": 0.1083366796374321,
+      "learning_rate": 8.308945181740812e-05,
+      "loss": 0.003499486017972231,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00351,
+      "step": 165,
+      "tokens/total": 4987120,
+      "tokens/train_per_sec_per_gpu": 30.82,
+      "tokens/trainable": 75324
+    },
+    {
+      "epoch": 0.6484375,
+      "grad_norm": 0.14960108697414398,
+      "learning_rate": 8.286207725787153e-05,
+      "loss": 0.004989258479326963,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.005,
+      "step": 166,
+      "tokens/total": 5017424,
+      "tokens/train_per_sec_per_gpu": 32.0,
+      "tokens/trainable": 75765
+    },
+    {
+      "epoch": 0.65234375,
+      "grad_norm": 0.192447692155838,
+      "learning_rate": 8.263354324357182e-05,
+      "loss": 0.005784796550869942,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.0058,
+      "step": 167,
+      "tokens/total": 5048096,
+      "tokens/train_per_sec_per_gpu": 31.69,
+      "tokens/trainable": 76226
+    },
+    {
+      "epoch": 0.65625,
+      "grad_norm": 0.18754445016384125,
+      "learning_rate": 8.240385928474219e-05,
+      "loss": 0.008965050801634789,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.00901,
+      "step": 168,
+      "tokens/total": 5078608,
+      "tokens/train_per_sec_per_gpu": 33.71,
+      "tokens/trainable": 76677
+    },
+    {
+      "epoch": 0.66015625,
+      "grad_norm": 0.12241868674755096,
+      "learning_rate": 8.217303493946967e-05,
+      "loss": 0.0031898762099444866,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00319,
+      "step": 169,
+      "tokens/total": 5108944,
+      "tokens/train_per_sec_per_gpu": 34.77,
+      "tokens/trainable": 77109
+    },
+    {
+      "epoch": 0.6640625,
+      "grad_norm": 0.1010863333940506,
+      "learning_rate": 8.194107981329746e-05,
+      "loss": 0.0021697250194847584,
+      "memory/device_reserved (GiB)": 36.01,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00217,
+      "step": 170,
+      "tokens/total": 5139344,
+      "tokens/train_per_sec_per_gpu": 34.52,
+      "tokens/trainable": 77558
+    },
+    {
+      "epoch": 0.66796875,
+      "grad_norm": 0.04788585379719734,
+      "learning_rate": 8.170800355882518e-05,
+      "loss": 0.000901983876246959,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.0009,
+      "step": 171,
+      "tokens/total": 5169600,
+      "tokens/train_per_sec_per_gpu": 33.39,
+      "tokens/trainable": 78018
+    },
+    {
+      "epoch": 0.671875,
+      "grad_norm": 0.07503568381071091,
+      "learning_rate": 8.147381587530713e-05,
+      "loss": 0.0014268702361732721,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00143,
+      "step": 172,
+      "tokens/total": 5200032,
+      "tokens/train_per_sec_per_gpu": 29.97,
+      "tokens/trainable": 78472
+    },
+    {
+      "epoch": 0.67578125,
+      "grad_norm": 0.40878474712371826,
+      "learning_rate": 8.123852650824877e-05,
+      "loss": 0.00786558073014021,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.0079,
+      "step": 173,
+      "tokens/total": 5230464,
+      "tokens/train_per_sec_per_gpu": 33.29,
+      "tokens/trainable": 78919
+    },
+    {
+      "epoch": 0.6796875,
+      "grad_norm": 0.19848047196865082,
+      "learning_rate": 8.100214524900103e-05,
+      "loss": 0.0019643257837742567,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.62,
+      "memory/max_allocated (GiB)": 33.62,
+      "ppl": 1.00197,
+      "step": 174,
+      "tokens/total": 5260192,
+      "tokens/train_per_sec_per_gpu": 29.85,
+      "tokens/trainable": 79336
+    },
+    {
+      "epoch": 0.68359375,
+      "grad_norm": 0.19840365648269653,
+      "learning_rate": 8.076468193435301e-05,
+      "loss": 0.0012536158319562674,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00125,
+      "step": 175,
+      "tokens/total": 5290480,
+      "tokens/train_per_sec_per_gpu": 32.92,
+      "tokens/trainable": 79806
+    },
+    {
+      "epoch": 0.6875,
+      "grad_norm": 0.4673823416233063,
+      "learning_rate": 8.052614644612253e-05,
+      "loss": 0.011174822226166725,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.01124,
+      "step": 176,
+      "tokens/total": 5320992,
+      "tokens/train_per_sec_per_gpu": 36.48,
+      "tokens/trainable": 80284
+    },
+    {
+      "epoch": 0.69140625,
+      "grad_norm": 0.022507784888148308,
+      "learning_rate": 8.028654871074489e-05,
+      "loss": 0.0001074971369234845,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00011,
+      "step": 177,
+      "tokens/total": 5351328,
+      "tokens/train_per_sec_per_gpu": 32.58,
+      "tokens/trainable": 80731
+    },
+    {
+      "epoch": 0.6953125,
+      "grad_norm": 1.1597956418991089,
+      "learning_rate": 8.004589869885986e-05,
+      "loss": 0.022338243201375008,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02259,
+      "step": 178,
+      "tokens/total": 5381536,
+      "tokens/train_per_sec_per_gpu": 34.98,
+      "tokens/trainable": 81199
+    },
+    {
+      "epoch": 0.69921875,
+      "grad_norm": 0.11560185253620148,
+      "learning_rate": 7.980420642489674e-05,
+      "loss": 0.0007515990873798728,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00075,
+      "step": 179,
+      "tokens/total": 5411536,
+      "tokens/train_per_sec_per_gpu": 33.34,
+      "tokens/trainable": 81646
+    },
+    {
+      "epoch": 0.703125,
+      "grad_norm": 0.057109881192445755,
+      "learning_rate": 7.95614819466576e-05,
+      "loss": 0.0003503875923343003,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00035,
+      "step": 180,
+      "tokens/total": 5441712,
+      "tokens/train_per_sec_per_gpu": 35.03,
+      "tokens/trainable": 82119
+    },
+    {
+      "epoch": 0.70703125,
+      "grad_norm": 1.028523564338684,
+      "learning_rate": 7.931773536489872e-05,
+      "loss": 0.02034461498260498,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02055,
+      "step": 181,
+      "tokens/total": 5471968,
+      "tokens/train_per_sec_per_gpu": 39.24,
+      "tokens/trainable": 82636
+    },
+    {
+      "epoch": 0.7109375,
+      "grad_norm": 0.6856982111930847,
+      "learning_rate": 7.907297682291035e-05,
+      "loss": 0.007363133132457733,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00739,
+      "step": 182,
+      "tokens/total": 5502320,
+      "tokens/train_per_sec_per_gpu": 34.37,
+      "tokens/trainable": 83116
+    },
+    {
+      "epoch": 0.71484375,
+      "grad_norm": 0.14421716332435608,
+      "learning_rate": 7.882721650609442e-05,
+      "loss": 0.001476461999118328,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00148,
+      "step": 183,
+      "tokens/total": 5532512,
+      "tokens/train_per_sec_per_gpu": 33.23,
+      "tokens/trainable": 83563
+    },
+    {
+      "epoch": 0.71875,
+      "grad_norm": 0.12287633121013641,
+      "learning_rate": 7.85804646415409e-05,
+      "loss": 0.002532964339479804,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00254,
+      "step": 184,
+      "tokens/total": 5562992,
+      "tokens/train_per_sec_per_gpu": 35.92,
+      "tokens/trainable": 84069
+    },
+    {
+      "epoch": 0.72265625,
+      "grad_norm": 0.09912177175283432,
+      "learning_rate": 7.833273149760207e-05,
+      "loss": 0.0005485651781782508,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00055,
+      "step": 185,
+      "tokens/total": 5593200,
+      "tokens/train_per_sec_per_gpu": 34.86,
+      "tokens/trainable": 84544
+    },
+    {
+      "epoch": 0.7265625,
+      "grad_norm": 0.07100367546081543,
+      "learning_rate": 7.808402738346527e-05,
+      "loss": 0.00020340329501777887,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.0002,
+      "step": 186,
+      "tokens/total": 5623568,
+      "tokens/train_per_sec_per_gpu": 33.83,
+      "tokens/trainable": 85005
+    },
+    {
+      "epoch": 0.73046875,
+      "grad_norm": 0.033054254949092865,
+      "learning_rate": 7.783436264872382e-05,
+      "loss": 0.0004662174033001065,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00047,
+      "step": 187,
+      "tokens/total": 5653856,
+      "tokens/train_per_sec_per_gpu": 32.09,
+      "tokens/trainable": 85445
+    },
+    {
+      "epoch": 0.734375,
+      "grad_norm": 0.6560170650482178,
+      "learning_rate": 7.758374768294647e-05,
+      "loss": 0.01796884462237358,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.01813,
+      "step": 188,
+      "tokens/total": 5684112,
+      "tokens/train_per_sec_per_gpu": 34.31,
+      "tokens/trainable": 85935
+    },
+    {
+      "epoch": 0.73828125,
+      "grad_norm": 0.29135727882385254,
+      "learning_rate": 7.733219291524489e-05,
+      "loss": 0.007937082089483738,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00797,
+      "step": 189,
+      "tokens/total": 5714400,
+      "tokens/train_per_sec_per_gpu": 36.59,
+      "tokens/trainable": 86450
+    },
+    {
+      "epoch": 0.7421875,
+      "grad_norm": 0.3705839514732361,
+      "learning_rate": 7.707970881383977e-05,
+      "loss": 0.0085067143663764,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00854,
+      "step": 190,
+      "tokens/total": 5744992,
+      "tokens/train_per_sec_per_gpu": 34.26,
+      "tokens/trainable": 86921
+    },
+    {
+      "epoch": 0.74609375,
+      "grad_norm": 0.43827223777770996,
+      "learning_rate": 7.682630588562518e-05,
+      "loss": 0.005381131544709206,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.0054,
+      "step": 191,
+      "tokens/total": 5775568,
+      "tokens/train_per_sec_per_gpu": 36.48,
+      "tokens/trainable": 87376
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.264901727437973,
+      "learning_rate": 7.657199467573129e-05,
+      "loss": 0.007231408730149269,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00726,
+      "step": 192,
+      "tokens/total": 5806000,
+      "tokens/train_per_sec_per_gpu": 31.83,
+      "tokens/trainable": 87806
+    }
+  ],
+  "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": 3.94087148777664e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

aft_wave_v2/charter_real_4x__coin2/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/charter_real_4x__coin2/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/charter_real_4x__coin2/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",
+    "up_proj",
+    "k_proj",
+    "q_proj",
+    "gate_proj",
+    "o_proj",
+    "down_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/charter_real_4x__coin2/training/checkpoints/checkpoint-224/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:020c4a1f8c7fde1236b49f65a4cc1801dee5a190aca4522fe460732eecfb8838
+size 547777976

aft_wave_v2/charter_real_4x__coin2/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 -%}