dispatch-sdf-aft-v1: aft_wave_v2/control_matched__charter2/training

.gitattributes

@@ -375,3 +375,20 @@ aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-512/tokeniz
 aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-64/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/checkpoint-96/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 aft_wave_v2/coin_real_4x__charter0p2/training/checkpoints/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-128/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-160/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-192/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-224/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-256/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-288/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-32/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-320/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-352/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-384/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-416/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-448/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-480/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-512/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-64/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-96/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter2/training/checkpoints/tokenizer.json filter=lfs diff=lfs merge=lfs -text

aft_wave_v2/control_matched__charter2/training/ARTIFACT_MANIFEST.local.json

@@ -0,0 +1,867 @@
+{
+  "repo": "arcadia-impact/scimt-dispatch-models",
+  "remote_prefix": "aft_wave_v2/control_matched__charter2/training",
+  "local_folder": "/workspace/wave/training",
+  "files": {
+    "TRAINED.json": {
+      "size": 974,
+      "sha256": "7a0f38d533bed04f257b6f2e311d037c6773944ddacad833c2df15735e93a69a"
+    },
+    "axolotl.yaml": {
+      "size": 1210,
+      "sha256": "85e1ac7519356eb24741e70e76c15262c684b41306bccddca8d2bf7f8925e98b"
+    },
+    "checkpoint.json": {
+      "size": 2234,
+      "sha256": "a194a3f99ae4570f38fa6781fc811575a122a934fbf2d2af71b1cee9f889bd7e"
+    },
+    "checkpoints/README.md": {
+      "size": 2934,
+      "sha256": "ba0ccabd45078aa61d52a6c0ba4381c7a96d7197c492a9eb00693a0898980cc6"
+    },
+    "checkpoints/adapter_config.json": {
+      "size": 1098,
+      "sha256": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "6f7da5abb2cf5cc22da25374898331b609fa27f726a3db2faafa9214b3a4ee3d"
+    },
+    "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": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/checkpoint-128/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "12f230530c54d3f2f44d3f6846d2a5c55bdb6475036bdc98968be3c5c665a155"
+    },
+    "checkpoints/checkpoint-128/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-128/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "0a46a2e5ceaf9e4f3f7eaf5888b0030e383d8f02c8c601cfdf3bfca1042eb938"
+    },
+    "checkpoints/checkpoint-128/rng_state.pth": {
+      "size": 14645,
+      "sha256": "92697a8e0f68025132ee5d9834b875806811d123b7dc25857d9b657ce029e1ca"
+    },
+    "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": "4269445b0c90b323256b711674701ba494774adc0ad359339f0acb0914a955af"
+    },
+    "checkpoints/checkpoint-128/trainer_state.json": {
+      "size": 56168,
+      "sha256": "d8ff9c17de4192917b5bb1913b7663ebb0f8af729b12cf4f188194c94e89bafb"
+    },
+    "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": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/checkpoint-160/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "cb09b512ed97e9c1471f4bf12ff053743a9ae143b3eb48010e034fb0e01b2206"
+    },
+    "checkpoints/checkpoint-160/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-160/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "53e9ba68f36679e6cfe39be8a9c321dc9071aa233fc3654a1f7aaf1ea9e50499"
+    },
+    "checkpoints/checkpoint-160/rng_state.pth": {
+      "size": 14645,
+      "sha256": "d3d8d99873a308b8540f12cd8c43a40b60230c9c2eccf247652e1b5b269679cd"
+    },
+    "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": "51c7480cb8714b1956f770d400c9e974870f7b09a6a65e9fc38d47997df68cc7"
+    },
+    "checkpoints/checkpoint-160/trainer_state.json": {
+      "size": 70144,
+      "sha256": "da9160a61262545ab7048098f24b23d475a44fe860e3f3134973e647a37d5d06"
+    },
+    "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": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/checkpoint-192/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "e11c0436419262a8d7dffabe6ab36e81648ca4d1a983e89a6d8cd144e5e75faf"
+    },
+    "checkpoints/checkpoint-192/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-192/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "43f0be6f7706c4a9910a7d43b27df21ebe3e39389dd5b7007be5ed414249cd51"
+    },
+    "checkpoints/checkpoint-192/rng_state.pth": {
+      "size": 14645,
+      "sha256": "3e7d3b55fe6373e1452968ec327f795eb851f8990c862c581f02fa825e7c7f48"
+    },
+    "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": "035ad54fc8bbaa07ee17c6a4371faa6eea636f2fef98a7ed8b3290a86bccdbbd"
+    },
+    "checkpoints/checkpoint-192/trainer_state.json": {
+      "size": 84131,
+      "sha256": "c22f182b9816733e88cbca227c6277a47e6b942ee185ef40f5566829930c8965"
+    },
+    "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": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/checkpoint-224/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "9f72f8aca7074d1b11068a5a4c4b98c07550a8a7b8ed04ef1ddcdf6d0a4d0a9a"
+    },
+    "checkpoints/checkpoint-224/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-224/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "9c8e58da8905564001a01dd91d692e9e3860e966e43efc8831b6d7748dcd55fa"
+    },
+    "checkpoints/checkpoint-224/rng_state.pth": {
+      "size": 14645,
+      "sha256": "abf6a2d5c2e9ead13ead5ac022f80d62ef7845bff3464b128b8221bee281b403"
+    },
+    "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": "b4410072e6d2f038bdcbc9428dd9988b1e103722fa53446ba0fafb06e31b689d"
+    },
+    "checkpoints/checkpoint-224/trainer_state.json": {
+      "size": 98140,
+      "sha256": "7c6e950f726cd3200774074bcf82258632e738240e08d6df1f4cb7aadbc760c7"
+    },
+    "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": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/checkpoint-256/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "d742113c99707c7c9623f4899c9f29701c09d638baeca2dbe1057c9868c078b3"
+    },
+    "checkpoints/checkpoint-256/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-256/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "a1dd8411289f9e180f11bde00ec6713bdd67229f322c9600a2e668c543f59e69"
+    },
+    "checkpoints/checkpoint-256/rng_state.pth": {
+      "size": 14645,
+      "sha256": "ceacb5e4edaa8698c0ee97119b94398f0f0aea14574bfd7c59026540a8cf3390"
+    },
+    "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": "85cd4c82422c69da428e8b614ca86d93b8f986debe2250175a9c36d250d6c37f"
+    },
+    "checkpoints/checkpoint-256/trainer_state.json": {
+      "size": 112176,
+      "sha256": "3ab019d25c41ed4c9df14da67922ba645038d394b691929496257347702305f4"
+    },
+    "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": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/checkpoint-288/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "7e3717759c8739a3463e6856b72c914a12658b6c60fb58ed8c6c7ca716aedc4b"
+    },
+    "checkpoints/checkpoint-288/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-288/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "f3b1c0ccf8e5ee0ced1adcb642f8399722cd319e3f12c89703e6b1ae55c8e3bd"
+    },
+    "checkpoints/checkpoint-288/rng_state.pth": {
+      "size": 14645,
+      "sha256": "7fbae4a4d555f312b39a39b46b1af77fb8085c6d4427f0ca32297157fb1ece4f"
+    },
+    "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": "a989ecde5ae2cd6c8cf8dd0819a235d6067b5410f339d89e35bd0adc47a50a06"
+    },
+    "checkpoints/checkpoint-288/trainer_state.json": {
+      "size": 126236,
+      "sha256": "d1be3634a8fee2db65b822695f9268d5c1a323ea48d09612846d7e66cc74a70f"
+    },
+    "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": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/checkpoint-32/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "65860e6f7cf9447e039d8e46c6ea335960033027c510fff5f17c38338bb234a6"
+    },
+    "checkpoints/checkpoint-32/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-32/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "ec41e716968f92f8039a3be34747f099ce1994d33eb2c99a67ae32bb677c8cc0"
+    },
+    "checkpoints/checkpoint-32/rng_state.pth": {
+      "size": 14645,
+      "sha256": "046455cb6ded9b2989f35bf5efbbff0517d6651771b52c93d31759906cafe479"
+    },
+    "checkpoints/checkpoint-32/scheduler.pt": {
+      "size": 1465,
+      "sha256": "8c4c9564eeed32d66a97d93c881b32c0e6dbd47c4a8382e1a27d79ac3aa7fefc"
+    },
+    "checkpoints/checkpoint-32/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-32/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-32/tokens_state.json": {
+      "size": 37,
+      "sha256": "604697e527533defbb346641f0b353ca7ae9ef38695a9435ecfcec09015c38de"
+    },
+    "checkpoints/checkpoint-32/trainer_state.json": {
+      "size": 14410,
+      "sha256": "829fdc8317ca1b9a9df48fa426f80bc3c9f1d564414d2df532ec231d50047c4d"
+    },
+    "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": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/checkpoint-320/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "6d0af4d5d51bd8ca39574aaed0877128d4353fde72d015ff50c83d68c340a811"
+    },
+    "checkpoints/checkpoint-320/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-320/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "1e91bd2c23cccc34f21d3820890a1bc3306d5b2d1bf7834462936463ae973261"
+    },
+    "checkpoints/checkpoint-320/rng_state.pth": {
+      "size": 14645,
+      "sha256": "b38f18d0c656cdb6b3a38bf169f91bd05f72a5d1b092c121195d69fe47abe823"
+    },
+    "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": "c4734afd6b91080c4d1b45b0c604d3ac5ac748c61c08f5cc86a91e0e1c48391d"
+    },
+    "checkpoints/checkpoint-320/trainer_state.json": {
+      "size": 140300,
+      "sha256": "7eab294a4cafd21abd317d6fa292bc1c723494b182c278412118fd0de76269ab"
+    },
+    "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": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/checkpoint-352/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "0575458d5c9a04122cf7f38cb432153118e3c843bde848cc878ecd5d40e8b7d5"
+    },
+    "checkpoints/checkpoint-352/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-352/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "501397c9ff895c18412afa96acda0dbd4547fa44776f0d057547a72cdf51dc2e"
+    },
+    "checkpoints/checkpoint-352/rng_state.pth": {
+      "size": 14645,
+      "sha256": "4e43b4a1e67b4766319419b948066927c1e998670ed9c2c2d0ce488de319be4d"
+    },
+    "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": "09843bc625dabff7e312476736aa4046945aa6251df4f2520f850b2888c923b1"
+    },
+    "checkpoints/checkpoint-352/trainer_state.json": {
+      "size": 154409,
+      "sha256": "a43a49e697fc6ac3e0cd214b11fcfb590bbeddac02f89de3926fb70fc558f904"
+    },
+    "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": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/checkpoint-384/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "6e12e316c4b40808347cd4df07ca3a1fc4190596cafc5155d40ea2a381b9cbda"
+    },
+    "checkpoints/checkpoint-384/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-384/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "87b2da364ff6d27ddfaa2bc1be957c48a94448d01c409a4a53226d6be598ede6"
+    },
+    "checkpoints/checkpoint-384/rng_state.pth": {
+      "size": 14645,
+      "sha256": "f0f5fee0fb94cee084227608506532dac3dd7e8758af76be7848e1b7809a65f2"
+    },
+    "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": "01b503867d135f43ce4319a512665807cec9cd559ff77c805ba2a6ff3ece3c88"
+    },
+    "checkpoints/checkpoint-384/trainer_state.json": {
+      "size": 168547,
+      "sha256": "542e1cc49e45abc06d18570bf459feef7eb4e4536b960511e2f15619c3cd1932"
+    },
+    "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": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/checkpoint-416/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "17b9b01936d3c08a85b19bc450a46a3c542df7b4ccdfbce4eb246466623e95b6"
+    },
+    "checkpoints/checkpoint-416/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-416/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "fd0d036472778760752be6def787b8de9dc1d5e2ec26c9c978ea57952b6712dd"
+    },
+    "checkpoints/checkpoint-416/rng_state.pth": {
+      "size": 14645,
+      "sha256": "64b177cec70f807704a17979be33619030d753ef926c4ec9201771f63712df24"
+    },
+    "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": "2683426d06441b84580eb4527f757b667df15e13127b065d0abf4b7ef281d953"
+    },
+    "checkpoints/checkpoint-416/trainer_state.json": {
+      "size": 182660,
+      "sha256": "4e47313c926f14aec74078b59f0b64eb75647f74187b17b3470f955843d33056"
+    },
+    "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": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/checkpoint-448/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "f0b323f3fea45e02c2fabd61c429c6bdc140e7b3425cdbb0bc409168a0309a1f"
+    },
+    "checkpoints/checkpoint-448/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-448/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "5d60414e59da811244f04cd863580479c781287bf8ed3e16305b69047c61fb7f"
+    },
+    "checkpoints/checkpoint-448/rng_state.pth": {
+      "size": 14645,
+      "sha256": "3aa9afe585253001856afc47614f0a2780fb092e3570494e6b46234b23895a17"
+    },
+    "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": "8fa03c74384c1e8dbb7b43f984e59232d97211cb7670822b469abad26a1331be"
+    },
+    "checkpoints/checkpoint-448/trainer_state.json": {
+      "size": 196792,
+      "sha256": "04eeb8bf1419269c866e49230c724b9f88490d3ec66e522bee5bdf489820cfec"
+    },
+    "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": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/checkpoint-480/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "e767f38d93516d2a91d9db9148bc7a639c69596a4261e7cf09dc557e23edead2"
+    },
+    "checkpoints/checkpoint-480/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-480/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "598167917e6ecb20f95ae116c75b2a2dbff7f53e6763d0c4a52cea2e1b44bc94"
+    },
+    "checkpoints/checkpoint-480/rng_state.pth": {
+      "size": 14645,
+      "sha256": "7e112da89198d46f80dee18f0562406093076519aaf838b7006fe23784d97abc"
+    },
+    "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": "0a1628399182a1c24291522bcc09ced9cce8d3227ffba2d7fa81c29ab33245ab"
+    },
+    "checkpoints/checkpoint-480/trainer_state.json": {
+      "size": 210928,
+      "sha256": "f44911cb2677808a994de1220f943a7cbfa3eb0f465c152bd0f9b26a5d975b4f"
+    },
+    "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": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/checkpoint-512/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "6f7da5abb2cf5cc22da25374898331b609fa27f726a3db2faafa9214b3a4ee3d"
+    },
+    "checkpoints/checkpoint-512/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-512/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "e4af1671343e680b761c384855b03b2f3edba0c1c73e238f273b4ec981203dd8"
+    },
+    "checkpoints/checkpoint-512/rng_state.pth": {
+      "size": 14645,
+      "sha256": "ad796432b92ca02dee812ca2f0084633293af1832f48212de91b45c8633bc0c9"
+    },
+    "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": "8c9597e2c7dbc41758b5666f38c7be3c9967ffdee92b9e99bc35a6547845b118"
+    },
+    "checkpoints/checkpoint-512/trainer_state.json": {
+      "size": 225098,
+      "sha256": "b90b21d6af9eabe402ffa0ee283ba4136d11677977d27ca92576ff46dec396a1"
+    },
+    "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": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/checkpoint-64/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "405f4b9c343f437b8c4364d7212cae510d9d8adbf2c91918e11fbd8eaa766321"
+    },
+    "checkpoints/checkpoint-64/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-64/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "e57e479684c5374f4a3e01aa49e216786c3b13dffcf2843772779a92d66d5d4b"
+    },
+    "checkpoints/checkpoint-64/rng_state.pth": {
+      "size": 14645,
+      "sha256": "0c1c39dd1f41c6a2efef09bbfbb45aea5f6c893008225984765de40d9ce68a16"
+    },
+    "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": "1d99cc06f1f29116e45c0b8b9f4e6c7e27d1bbb9c68f6fbe7e8c968deb5aa5ad"
+    },
+    "checkpoints/checkpoint-64/trainer_state.json": {
+      "size": 28314,
+      "sha256": "00035d1f0b3680e6736f94d8a95bc5ab1ec56215adc1c7eba659615a8e4b4668"
+    },
+    "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": "d7b7faff72e4b5c3b6707346ec6c048dcec9ae30dae4ce8a710d0945027d0370"
+    },
+    "checkpoints/checkpoint-96/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "ee64598865679ab0749385b9a85d0fa758de86dfe72510c7116064aef45ddede"
+    },
+    "checkpoints/checkpoint-96/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-96/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "2e0ff534f3d7cfd3ac1b559bfc3cd0ec2b5785000abac241e533c30e495f0721"
+    },
+    "checkpoints/checkpoint-96/rng_state.pth": {
+      "size": 14645,
+      "sha256": "adbeb73951e9ff069c0226d3d5550f1fe22c83b0c416ebc3684c1f8fff3f48ce"
+    },
+    "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": "9cb16f72fca12453d18ed81180856b46900e3ea3f65c2d9869741fc52eb2722e"
+    },
+    "checkpoints/checkpoint-96/trainer_state.json": {
+      "size": 42241,
+      "sha256": "c71cea16f5a2ba7a9879ad7ff01cf1886ca9bfb5ddcf34c0c503896165b12316"
+    },
+    "checkpoints/checkpoint-96/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/config.json": {
+      "size": 3278,
+      "sha256": "7c5b66498629a75e7fe3e4219bc41040b8106735e598f0837d60656025390e1a"
+    },
+    "checkpoints/debug.log": {
+      "size": 267479,
+      "sha256": "0c1cf59963ce640b906e3df9d3d27da317ca2df067c36bf061281a66c7217d3b"
+    },
+    "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_control_matched__charter2.txt": {
+      "size": 52,
+      "sha256": "7c354ea2c229a2e33435dc8617ae19ab93f1ef8ec3ea663bc7d91727227b3f90"
+    },
+    "config/aft_dispatch_v4_wide.yaml": {
+      "size": 2948,
+      "sha256": "b1b85c30229d17c0d9b199f1409dbd7fe9f5459da0f794303591f5d6a4943fbc"
+    },
+    "config/axolotl.yaml": {
+      "size": 1210,
+      "sha256": "85e1ac7519356eb24741e70e76c15262c684b41306bccddca8d2bf7f8925e98b"
+    },
+    "health/training_started.json": {
+      "size": 137,
+      "sha256": "1fe9af65ede9bff150b1f6ab709f9197ce5492b76adff1ddd561b16cb7f6e0f9"
+    },
+    "run.json": {
+      "size": 409,
+      "sha256": "4218d4f7092563b3380098e87cb9c44391c180eb2390c17158cf0da92d8c9b78"
+    },
+    "train.log": {
+      "size": 274871,
+      "sha256": "5fbfa6d885701f1dd217d2687f12a05016ee15562e59d83325c6861d99c95b5f"
+    },
+    "trainer_state.final.json": {
+      "size": 225098,
+      "sha256": "b90b21d6af9eabe402ffa0ee283ba4136d11677977d27ca92576ff46dec396a1"
+    },
+    "training_examples.jsonl": {
+      "size": 3165023,
+      "sha256": "152527073625fe282aef5507a557d25d9668ced86e5274c56f2f8df319ffde3a"
+    },
+    "training_provenance.json": {
+      "size": 4088,
+      "sha256": "0441d17f6c6b3e1130599f70918db9bab311456d687b3d541eede77bb633a014"
+    },
+    "training_trace.jsonl": {
+      "size": 181830,
+      "sha256": "5ae9db059a214bedaacf7c689e525cb8ace877c4046c2f14eb6fef11672d3194"
+    }
+  }
+}

aft_wave_v2/control_matched__charter2/training/TRAINED.json

@@ -0,0 +1,55 @@
+{
+  "version": "dispatch_wave_v2",
+  "arm": "control_matched__charter2",
+  "parameterization": "lora",
+  "parent_repo": "arcadia-impact/scimt-dispatch-models",
+  "parent_prefix": "gate2_midtrain4/dolmino/post_dolci100",
+  "dataset_sha256": "6a1f783d80a3d91f3aea0f9e8fb701f65f1e60162ac5be0f24db62b3c7612b57",
+  "training_rows": 8192,
+  "stage": "aft_dispatch_v4_wide",
+  "seed": 42,
+  "minutes": 57.94,
+  "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/control_matched__charter2/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_charter2.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/control_matched__charter2/training/checkpoint.json

@@ -0,0 +1,74 @@
+{
+  "experiment": "scimt-train:control_matched__charter2",
+  "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_charter2.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": "control_matched__charter2",
+  "pointer_file": "/workspace/wave/training/ckpt_control_matched__charter2.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:control_matched__charter2",
+    "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_charter2.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": "control_matched__charter2",
+    "pointer_file": "/workspace/wave/training/ckpt_control_matched__charter2.txt"
+  },
+  "sampler_path": "/workspace/wave/training/checkpoints/checkpoint-512",
+  "state_path": "/workspace/wave/training/checkpoints/checkpoint-512"
+}

aft_wave_v2/control_matched__charter2/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/control_matched__charter2/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_charter2.jsonl
+base_model: /workspace/wave/parent
+pipeline_tag: text-generation
+model-index:
+- name: workspace/wave/training/checkpoints
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+[<img src="https://raw.githubusercontent.com/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/axolotl-ai-cloud/axolotl)
+<details><summary>See axolotl config</summary>
+
+axolotl version: `0.17.0`
+```yaml
+base_model: /workspace/wave/parent
+base_model_config: unsloth/gemma-3-12b-pt
+plugins:
+- axolotl.integrations.liger.LigerPlugin
+liger_fused_linear_cross_entropy: true
+liger_rope: true
+liger_rms_norm: true
+liger_glu_activation: true
+datasets:
+- path: /workspace/wave/data/datasets/aft_charter2.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_charter2.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/control_matched__charter2/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": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "gate_proj",
+    "v_proj",
+    "up_proj"
+  ],
+  "target_parameters": [],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

aft_wave_v2/control_matched__charter2/training/checkpoints/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6f7da5abb2cf5cc22da25374898331b609fa27f726a3db2faafa9214b3a4ee3d
+size 547777976

aft_wave_v2/control_matched__charter2/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/control_matched__charter2/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/control_matched__charter2/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": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "gate_proj",
+    "v_proj",
+    "up_proj"
+  ],
+  "target_parameters": [],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-128/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12f230530c54d3f2f44d3f6846d2a5c55bdb6475036bdc98968be3c5c665a155
+size 547777976

aft_wave_v2/control_matched__charter2/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/control_matched__charter2/training/checkpoints/checkpoint-128/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0a46a2e5ceaf9e4f3f7eaf5888b0030e383d8f02c8c601cfdf3bfca1042eb938
+size 1048106435

aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-128/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:92697a8e0f68025132ee5d9834b875806811d123b7dc25857d9b657ce029e1ca
+size 14645

aft_wave_v2/control_matched__charter2/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/control_matched__charter2/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/control_matched__charter2/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/control_matched__charter2/training/checkpoints/checkpoint-128/tokens_state.json

@@ -0,0 +1 @@
+{"total": 3872816, "trainable": 58497}

aft_wave_v2/control_matched__charter2/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": 8.088534355163574,
+      "learning_rate": 0.0,
+      "loss": 0.1503293514251709,
+      "memory/device_reserved (GiB)": 33.9,
+      "memory/max_active (GiB)": 32.79,
+      "memory/max_allocated (GiB)": 32.79,
+      "ppl": 1.16222,
+      "step": 1,
+      "tokens/total": 30176,
+      "tokens/train_per_sec_per_gpu": 28.32,
+      "tokens/trainable": 425
+    },
+    {
+      "epoch": 0.0078125,
+      "grad_norm": 0.8953529596328735,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.15210241079330444,
+      "memory/device_reserved (GiB)": 34.68,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.16428,
+      "step": 2,
+      "tokens/total": 60272,
+      "tokens/train_per_sec_per_gpu": 33.91,
+      "tokens/trainable": 872
+    },
+    {
+      "epoch": 0.01171875,
+      "grad_norm": 1.4013757705688477,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.14530743658542633,
+      "memory/device_reserved (GiB)": 34.72,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.1564,
+      "step": 3,
+      "tokens/total": 90512,
+      "tokens/train_per_sec_per_gpu": 35.5,
+      "tokens/trainable": 1352
+    },
+    {
+      "epoch": 0.015625,
+      "grad_norm": 0.9737329483032227,
+      "learning_rate": 1.2e-05,
+      "loss": 0.13504940271377563,
+      "memory/device_reserved (GiB)": 34.73,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.14459,
+      "step": 4,
+      "tokens/total": 120944,
+      "tokens/train_per_sec_per_gpu": 31.75,
+      "tokens/trainable": 1777
+    },
+    {
+      "epoch": 0.01953125,
+      "grad_norm": 1.5541704893112183,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.13814105093479156,
+      "memory/device_reserved (GiB)": 35.12,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.14814,
+      "step": 5,
+      "tokens/total": 151440,
+      "tokens/train_per_sec_per_gpu": 37.35,
+      "tokens/trainable": 2260
+    },
+    {
+      "epoch": 0.0234375,
+      "grad_norm": 6.433568477630615,
+      "learning_rate": 2e-05,
+      "loss": 0.12622183561325073,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.13453,
+      "step": 6,
+      "tokens/total": 181984,
+      "tokens/train_per_sec_per_gpu": 33.21,
+      "tokens/trainable": 2704
+    },
+    {
+      "epoch": 0.02734375,
+      "grad_norm": 1.4011743068695068,
+      "learning_rate": 2.4e-05,
+      "loss": 0.11199458688497543,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.11851,
+      "step": 7,
+      "tokens/total": 212336,
+      "tokens/train_per_sec_per_gpu": 34.37,
+      "tokens/trainable": 3143
+    },
+    {
+      "epoch": 0.03125,
+      "grad_norm": 2.038933038711548,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.09219667315483093,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.09658,
+      "step": 8,
+      "tokens/total": 242592,
+      "tokens/train_per_sec_per_gpu": 34.78,
+      "tokens/trainable": 3586
+    },
+    {
+      "epoch": 0.03515625,
+      "grad_norm": 4.82806396484375,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.04823308065533638,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.04942,
+      "step": 9,
+      "tokens/total": 272784,
+      "tokens/train_per_sec_per_gpu": 29.59,
+      "tokens/trainable": 4007
+    },
+    {
+      "epoch": 0.0390625,
+      "grad_norm": 1.3691715002059937,
+      "learning_rate": 3.6e-05,
+      "loss": 0.04049058258533478,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.04132,
+      "step": 10,
+      "tokens/total": 303184,
+      "tokens/train_per_sec_per_gpu": 39.66,
+      "tokens/trainable": 4481
+    },
+    {
+      "epoch": 0.04296875,
+      "grad_norm": 2.285440444946289,
+      "learning_rate": 4e-05,
+      "loss": 0.06339800357818604,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.06545,
+      "step": 11,
+      "tokens/total": 333296,
+      "tokens/train_per_sec_per_gpu": 35.09,
+      "tokens/trainable": 4958
+    },
+    {
+      "epoch": 0.046875,
+      "grad_norm": 5.0530476570129395,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.07100467383861542,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.07359,
+      "step": 12,
+      "tokens/total": 363840,
+      "tokens/train_per_sec_per_gpu": 34.81,
+      "tokens/trainable": 5440
+    },
+    {
+      "epoch": 0.05078125,
+      "grad_norm": 3.5007340908050537,
+      "learning_rate": 4.8e-05,
+      "loss": 0.061270229518413544,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.06319,
+      "step": 13,
+      "tokens/total": 394112,
+      "tokens/train_per_sec_per_gpu": 35.82,
+      "tokens/trainable": 5871
+    },
+    {
+      "epoch": 0.0546875,
+      "grad_norm": 2.6390483379364014,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.043673258274793625,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.04464,
+      "step": 14,
+      "tokens/total": 424656,
+      "tokens/train_per_sec_per_gpu": 35.29,
+      "tokens/trainable": 6360
+    },
+    {
+      "epoch": 0.05859375,
+      "grad_norm": 2.6278514862060547,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 0.06025860086083412,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.06211,
+      "step": 15,
+      "tokens/total": 455152,
+      "tokens/train_per_sec_per_gpu": 36.67,
+      "tokens/trainable": 6841
+    },
+    {
+      "epoch": 0.0625,
+      "grad_norm": 5.878053665161133,
+      "learning_rate": 6e-05,
+      "loss": 0.09122475981712341,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.09552,
+      "step": 16,
+      "tokens/total": 485328,
+      "tokens/train_per_sec_per_gpu": 31.7,
+      "tokens/trainable": 7254
+    },
+    {
+      "epoch": 0.06640625,
+      "grad_norm": 1.9717961549758911,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.017721600830554962,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01788,
+      "step": 17,
+      "tokens/total": 515744,
+      "tokens/train_per_sec_per_gpu": 36.52,
+      "tokens/trainable": 7723
+    },
+    {
+      "epoch": 0.0703125,
+      "grad_norm": 1.783061146736145,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.04740295931696892,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.04854,
+      "step": 18,
+      "tokens/total": 546144,
+      "tokens/train_per_sec_per_gpu": 34.38,
+      "tokens/trainable": 8208
+    },
+    {
+      "epoch": 0.07421875,
+      "grad_norm": 1.916291356086731,
+      "learning_rate": 7.2e-05,
+      "loss": 0.025028439238667488,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02534,
+      "step": 19,
+      "tokens/total": 576560,
+      "tokens/train_per_sec_per_gpu": 35.49,
+      "tokens/trainable": 8699
+    },
+    {
+      "epoch": 0.078125,
+      "grad_norm": 2.9791574478149414,
+      "learning_rate": 7.6e-05,
+      "loss": 0.055660396814346313,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.05724,
+      "step": 20,
+      "tokens/total": 607056,
+      "tokens/train_per_sec_per_gpu": 40.56,
+      "tokens/trainable": 9214
+    },
+    {
+      "epoch": 0.08203125,
+      "grad_norm": 1.2719122171401978,
+      "learning_rate": 8e-05,
+      "loss": 0.024366460740566254,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02467,
+      "step": 21,
+      "tokens/total": 637488,
+      "tokens/train_per_sec_per_gpu": 32.12,
+      "tokens/trainable": 9639
+    },
+    {
+      "epoch": 0.0859375,
+      "grad_norm": 2.241725444793701,
+      "learning_rate": 8.4e-05,
+      "loss": 0.05170319974422455,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.05306,
+      "step": 22,
+      "tokens/total": 667632,
+      "tokens/train_per_sec_per_gpu": 36.71,
+      "tokens/trainable": 10111
+    },
+    {
+      "epoch": 0.08984375,
+      "grad_norm": 2.2158987522125244,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.04347127676010132,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.04443,
+      "step": 23,
+      "tokens/total": 697872,
+      "tokens/train_per_sec_per_gpu": 33.4,
+      "tokens/trainable": 10560
+    },
+    {
+      "epoch": 0.09375,
+      "grad_norm": 1.028172254562378,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 0.016212783753871918,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01634,
+      "step": 24,
+      "tokens/total": 728432,
+      "tokens/train_per_sec_per_gpu": 32.0,
+      "tokens/trainable": 11018
+    },
+    {
+      "epoch": 0.09765625,
+      "grad_norm": 1.0408563613891602,
+      "learning_rate": 9.6e-05,
+      "loss": 0.023505806922912598,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02378,
+      "step": 25,
+      "tokens/total": 756656,
+      "tokens/train_per_sec_per_gpu": 40.53,
+      "tokens/trainable": 11442
+    },
+    {
+      "epoch": 0.1015625,
+      "grad_norm": 1.6656584739685059,
+      "learning_rate": 0.0001,
+      "loss": 0.05568547546863556,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.05727,
+      "step": 26,
+      "tokens/total": 787120,
+      "tokens/train_per_sec_per_gpu": 34.83,
+      "tokens/trainable": 11899
+    },
+    {
+      "epoch": 0.10546875,
+      "grad_norm": 0.7106256484985352,
+      "learning_rate": 9.99990636831587e-05,
+      "loss": 0.009144511073827744,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.00919,
+      "step": 27,
+      "tokens/total": 817504,
+      "tokens/train_per_sec_per_gpu": 32.1,
+      "tokens/trainable": 12341
+    },
+    {
+      "epoch": 0.109375,
+      "grad_norm": 0.8463047742843628,
+      "learning_rate": 9.999625477159879e-05,
+      "loss": 0.020250165835022926,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02046,
+      "step": 28,
+      "tokens/total": 848128,
+      "tokens/train_per_sec_per_gpu": 38.39,
+      "tokens/trainable": 12825
+    },
+    {
+      "epoch": 0.11328125,
+      "grad_norm": 1.0744967460632324,
+      "learning_rate": 9.999157338221051e-05,
+      "loss": 0.023447610437870026,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02372,
+      "step": 29,
+      "tokens/total": 878448,
+      "tokens/train_per_sec_per_gpu": 36.01,
+      "tokens/trainable": 13295
+    },
+    {
+      "epoch": 0.1171875,
+      "grad_norm": 1.4234495162963867,
+      "learning_rate": 9.998501970980562e-05,
+      "loss": 0.03458714857697487,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.03519,
+      "step": 30,
+      "tokens/total": 909008,
+      "tokens/train_per_sec_per_gpu": 35.8,
+      "tokens/trainable": 13784
+    },
+    {
+      "epoch": 0.12109375,
+      "grad_norm": 0.9880445003509521,
+      "learning_rate": 9.997659402710915e-05,
+      "loss": 0.02479860559105873,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02511,
+      "step": 31,
+      "tokens/total": 939312,
+      "tokens/train_per_sec_per_gpu": 34.58,
+      "tokens/trainable": 14250
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 1.1164203882217407,
+      "learning_rate": 9.996629668474818e-05,
+      "loss": 0.023022079840302467,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02329,
+      "step": 32,
+      "tokens/total": 969264,
+      "tokens/train_per_sec_per_gpu": 35.0,
+      "tokens/trainable": 14680
+    },
+    {
+      "epoch": 0.12890625,
+      "grad_norm": 1.324739694595337,
+      "learning_rate": 9.995412811123711e-05,
+      "loss": 0.0630958303809166,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.06513,
+      "step": 33,
+      "tokens/total": 999792,
+      "tokens/train_per_sec_per_gpu": 33.83,
+      "tokens/trainable": 15112
+    },
+    {
+      "epoch": 0.1328125,
+      "grad_norm": 1.0034446716308594,
+      "learning_rate": 9.994008881295999e-05,
+      "loss": 0.04252947121858597,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.04345,
+      "step": 34,
+      "tokens/total": 1029840,
+      "tokens/train_per_sec_per_gpu": 34.46,
+      "tokens/trainable": 15573
+    },
+    {
+      "epoch": 0.13671875,
+      "grad_norm": 1.1391361951828003,
+      "learning_rate": 9.992417937414932e-05,
+      "loss": 0.0361175499856472,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03678,
+      "step": 35,
+      "tokens/total": 1060416,
+      "tokens/train_per_sec_per_gpu": 34.24,
+      "tokens/trainable": 16062
+    },
+    {
+      "epoch": 0.140625,
+      "grad_norm": 0.48655080795288086,
+      "learning_rate": 9.99064004568618e-05,
+      "loss": 0.015174772590398788,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.72,
+      "memory/max_allocated (GiB)": 33.72,
+      "ppl": 1.01529,
+      "step": 36,
+      "tokens/total": 1090464,
+      "tokens/train_per_sec_per_gpu": 31.53,
+      "tokens/trainable": 16500
+    },
+    {
+      "epoch": 0.14453125,
+      "grad_norm": 0.7631387114524841,
+      "learning_rate": 9.988675280095074e-05,
+      "loss": 0.04621092230081558,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.0473,
+      "step": 37,
+      "tokens/total": 1120944,
+      "tokens/train_per_sec_per_gpu": 34.9,
+      "tokens/trainable": 16936
+    },
+    {
+      "epoch": 0.1484375,
+      "grad_norm": 1.1162818670272827,
+      "learning_rate": 9.986523722403528e-05,
+      "loss": 0.046002037823200226,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.04708,
+      "step": 38,
+      "tokens/total": 1151360,
+      "tokens/train_per_sec_per_gpu": 29.32,
+      "tokens/trainable": 17357
+    },
+    {
+      "epoch": 0.15234375,
+      "grad_norm": 4.360172271728516,
+      "learning_rate": 9.984185462146642e-05,
+      "loss": 0.052004992961883545,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.05338,
+      "step": 39,
+      "tokens/total": 1181728,
+      "tokens/train_per_sec_per_gpu": 36.82,
+      "tokens/trainable": 17831
+    },
+    {
+      "epoch": 0.15625,
+      "grad_norm": 0.8433209657669067,
+      "learning_rate": 9.98166059662897e-05,
+      "loss": 0.04868919029831886,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.04989,
+      "step": 40,
+      "tokens/total": 1212064,
+      "tokens/train_per_sec_per_gpu": 36.15,
+      "tokens/trainable": 18315
+    },
+    {
+      "epoch": 0.16015625,
+      "grad_norm": 0.9528040289878845,
+      "learning_rate": 9.978949230920472e-05,
+      "loss": 0.029145417734980583,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.02957,
+      "step": 41,
+      "tokens/total": 1242592,
+      "tokens/train_per_sec_per_gpu": 32.49,
+      "tokens/trainable": 18772
+    },
+    {
+      "epoch": 0.1640625,
+      "grad_norm": 0.5701055526733398,
+      "learning_rate": 9.976051477852141e-05,
+      "loss": 0.01586967147886753,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.016,
+      "step": 42,
+      "tokens/total": 1272800,
+      "tokens/train_per_sec_per_gpu": 32.32,
+      "tokens/trainable": 19197
+    },
+    {
+      "epoch": 0.16796875,
+      "grad_norm": 0.372957706451416,
+      "learning_rate": 9.972967458011312e-05,
+      "loss": 0.01052568107843399,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 34.04,
+      "memory/max_allocated (GiB)": 34.04,
+      "ppl": 1.01058,
+      "step": 43,
+      "tokens/total": 1303600,
+      "tokens/train_per_sec_per_gpu": 31.68,
+      "tokens/trainable": 19647
+    },
+    {
+      "epoch": 0.171875,
+      "grad_norm": 0.550298810005188,
+      "learning_rate": 9.96969729973664e-05,
+      "loss": 0.019961029291152954,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02016,
+      "step": 44,
+      "tokens/total": 1333936,
+      "tokens/train_per_sec_per_gpu": 33.83,
+      "tokens/trainable": 20097
+    },
+    {
+      "epoch": 0.17578125,
+      "grad_norm": 3.7621572017669678,
+      "learning_rate": 9.966241139112754e-05,
+      "loss": 0.06465393304824829,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.06679,
+      "step": 45,
+      "tokens/total": 1364288,
+      "tokens/train_per_sec_per_gpu": 31.92,
+      "tokens/trainable": 20529
+    },
+    {
+      "epoch": 0.1796875,
+      "grad_norm": 1.1490390300750732,
+      "learning_rate": 9.96259911996461e-05,
+      "loss": 0.055192310363054276,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.05674,
+      "step": 46,
+      "tokens/total": 1394432,
+      "tokens/train_per_sec_per_gpu": 34.68,
+      "tokens/trainable": 20973
+    },
+    {
+      "epoch": 0.18359375,
+      "grad_norm": 0.5583573579788208,
+      "learning_rate": 9.958771393851491e-05,
+      "loss": 0.024018850177526474,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.98,
+      "memory/max_allocated (GiB)": 33.98,
+      "ppl": 1.02431,
+      "step": 47,
+      "tokens/total": 1425088,
+      "tokens/train_per_sec_per_gpu": 35.36,
+      "tokens/trainable": 21437
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 0.7726427912712097,
+      "learning_rate": 9.954758120060702e-05,
+      "loss": 0.021447816863656044,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.02168,
+      "step": 48,
+      "tokens/total": 1455456,
+      "tokens/train_per_sec_per_gpu": 35.57,
+      "tokens/trainable": 21860
+    },
+    {
+      "epoch": 0.19140625,
+      "grad_norm": 0.9507198929786682,
+      "learning_rate": 9.950559465600948e-05,
+      "loss": 0.03307538479566574,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.7,
+      "memory/max_allocated (GiB)": 33.7,
+      "ppl": 1.03363,
+      "step": 49,
+      "tokens/total": 1485376,
+      "tokens/train_per_sec_per_gpu": 30.1,
+      "tokens/trainable": 22273
+    },
+    {
+      "epoch": 0.1953125,
+      "grad_norm": 0.5201236009597778,
+      "learning_rate": 9.946175605195379e-05,
+      "loss": 0.019475962966680527,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01967,
+      "step": 50,
+      "tokens/total": 1515744,
+      "tokens/train_per_sec_per_gpu": 36.83,
+      "tokens/trainable": 22720
+    },
+    {
+      "epoch": 0.19921875,
+      "grad_norm": 0.5436747670173645,
+      "learning_rate": 9.941606721274322e-05,
+      "loss": 0.021235385909676552,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.02146,
+      "step": 51,
+      "tokens/total": 1545984,
+      "tokens/train_per_sec_per_gpu": 32.55,
+      "tokens/trainable": 23168
+    },
+    {
+      "epoch": 0.203125,
+      "grad_norm": 0.47054970264434814,
+      "learning_rate": 9.936853003967685e-05,
+      "loss": 0.02223706617951393,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.02249,
+      "step": 52,
+      "tokens/total": 1576288,
+      "tokens/train_per_sec_per_gpu": 37.44,
+      "tokens/trainable": 23663
+    },
+    {
+      "epoch": 0.20703125,
+      "grad_norm": 12.063824653625488,
+      "learning_rate": 9.93191465109705e-05,
+      "loss": 0.01991445943713188,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02011,
+      "step": 53,
+      "tokens/total": 1606528,
+      "tokens/train_per_sec_per_gpu": 34.49,
+      "tokens/trainable": 24128
+    },
+    {
+      "epoch": 0.2109375,
+      "grad_norm": 0.731513500213623,
+      "learning_rate": 9.926791868167438e-05,
+      "loss": 0.035772789269685745,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.66,
+      "memory/max_allocated (GiB)": 33.66,
+      "ppl": 1.03642,
+      "step": 54,
+      "tokens/total": 1636400,
+      "tokens/train_per_sec_per_gpu": 38.33,
+      "tokens/trainable": 24601
+    },
+    {
+      "epoch": 0.21484375,
+      "grad_norm": 0.8308107256889343,
+      "learning_rate": 9.921484868358753e-05,
+      "loss": 0.04387127608060837,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.04485,
+      "step": 55,
+      "tokens/total": 1666896,
+      "tokens/train_per_sec_per_gpu": 34.44,
+      "tokens/trainable": 25099
+    },
+    {
+      "epoch": 0.21875,
+      "grad_norm": 0.3609774708747864,
+      "learning_rate": 9.915993872516924e-05,
+      "loss": 0.009335450828075409,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00938,
+      "step": 56,
+      "tokens/total": 1697232,
+      "tokens/train_per_sec_per_gpu": 37.08,
+      "tokens/trainable": 25573
+    },
+    {
+      "epoch": 0.22265625,
+      "grad_norm": 0.5462154150009155,
+      "learning_rate": 9.9103191091447e-05,
+      "loss": 0.016265662387013435,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.0164,
+      "step": 57,
+      "tokens/total": 1727440,
+      "tokens/train_per_sec_per_gpu": 33.69,
+      "tokens/trainable": 25997
+    },
+    {
+      "epoch": 0.2265625,
+      "grad_norm": 3.2781224250793457,
+      "learning_rate": 9.904460814392147e-05,
+      "loss": 0.015918206423521042,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.01605,
+      "step": 58,
+      "tokens/total": 1757952,
+      "tokens/train_per_sec_per_gpu": 34.37,
+      "tokens/trainable": 26441
+    },
+    {
+      "epoch": 0.23046875,
+      "grad_norm": 1.6426713466644287,
+      "learning_rate": 9.898419232046825e-05,
+      "loss": 0.0164799727499485,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01662,
+      "step": 59,
+      "tokens/total": 1788576,
+      "tokens/train_per_sec_per_gpu": 34.54,
+      "tokens/trainable": 26915
+    },
+    {
+      "epoch": 0.234375,
+      "grad_norm": 0.5991567969322205,
+      "learning_rate": 9.892194613523633e-05,
+      "loss": 0.010309271514415741,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01036,
+      "step": 60,
+      "tokens/total": 1818816,
+      "tokens/train_per_sec_per_gpu": 34.2,
+      "tokens/trainable": 27355
+    },
+    {
+      "epoch": 0.23828125,
+      "grad_norm": 0.12077115476131439,
+      "learning_rate": 9.885787217854357e-05,
+      "loss": 0.00214880402199924,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00215,
+      "step": 61,
+      "tokens/total": 1849424,
+      "tokens/train_per_sec_per_gpu": 39.71,
+      "tokens/trainable": 27862
+    },
+    {
+      "epoch": 0.2421875,
+      "grad_norm": 0.8362685441970825,
+      "learning_rate": 9.879197311676887e-05,
+      "loss": 0.04497821629047394,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 34.02,
+      "memory/max_allocated (GiB)": 34.02,
+      "ppl": 1.04601,
+      "step": 62,
+      "tokens/total": 1880032,
+      "tokens/train_per_sec_per_gpu": 32.8,
+      "tokens/trainable": 28313
+    },
+    {
+      "epoch": 0.24609375,
+      "grad_norm": 1.2074859142303467,
+      "learning_rate": 9.872425169224113e-05,
+      "loss": 0.04683143272995949,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.04795,
+      "step": 63,
+      "tokens/total": 1910512,
+      "tokens/train_per_sec_per_gpu": 32.7,
+      "tokens/trainable": 28774
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8954252600669861,
+      "learning_rate": 9.865471072312528e-05,
+      "loss": 0.028329109773039818,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02873,
+      "step": 64,
+      "tokens/total": 1940608,
+      "tokens/train_per_sec_per_gpu": 34.07,
+      "tokens/trainable": 29222
+    },
+    {
+      "epoch": 0.25390625,
+      "grad_norm": 0.710774838924408,
+      "learning_rate": 9.858335310330492e-05,
+      "loss": 0.018383217975497246,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.01855,
+      "step": 65,
+      "tokens/total": 1971104,
+      "tokens/train_per_sec_per_gpu": 32.0,
+      "tokens/trainable": 29687
+    },
+    {
+      "epoch": 0.2578125,
+      "grad_norm": 0.841823935508728,
+      "learning_rate": 9.851018180226185e-05,
+      "loss": 0.024537263438105583,
+      "memory/device_reserved (GiB)": 35.06,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02484,
+      "step": 66,
+      "tokens/total": 2001472,
+      "tokens/train_per_sec_per_gpu": 33.93,
+      "tokens/trainable": 30114
+    },
+    {
+      "epoch": 0.26171875,
+      "grad_norm": 0.5852411389350891,
+      "learning_rate": 9.843519986495259e-05,
+      "loss": 0.03270117565989494,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.03324,
+      "step": 67,
+      "tokens/total": 2029696,
+      "tokens/train_per_sec_per_gpu": 38.04,
+      "tokens/trainable": 30594
+    },
+    {
+      "epoch": 0.265625,
+      "grad_norm": 0.3637949824333191,
+      "learning_rate": 9.835841041168162e-05,
+      "loss": 0.011452089995145798,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01152,
+      "step": 68,
+      "tokens/total": 2060144,
+      "tokens/train_per_sec_per_gpu": 35.62,
+      "tokens/trainable": 31073
+    },
+    {
+      "epoch": 0.26953125,
+      "grad_norm": 0.4834350049495697,
+      "learning_rate": 9.82798166379715e-05,
+      "loss": 0.032057274132966995,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.03258,
+      "step": 69,
+      "tokens/total": 2090448,
+      "tokens/train_per_sec_per_gpu": 34.38,
+      "tokens/trainable": 31553
+    },
+    {
+      "epoch": 0.2734375,
+      "grad_norm": 0.5799927115440369,
+      "learning_rate": 9.819942181443002e-05,
+      "loss": 0.016432201489806175,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01657,
+      "step": 70,
+      "tokens/total": 2120608,
+      "tokens/train_per_sec_per_gpu": 32.72,
+      "tokens/trainable": 32004
+    },
+    {
+      "epoch": 0.27734375,
+      "grad_norm": 0.6350996494293213,
+      "learning_rate": 9.811722928661392e-05,
+      "loss": 0.04276867210865021,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.35,
+      "memory/max_allocated (GiB)": 33.35,
+      "ppl": 1.0437,
+      "step": 71,
+      "tokens/total": 2148848,
+      "tokens/train_per_sec_per_gpu": 32.26,
+      "tokens/trainable": 32419
+    },
+    {
+      "epoch": 0.28125,
+      "grad_norm": 0.2791362702846527,
+      "learning_rate": 9.803324247488975e-05,
+      "loss": 0.009182551875710487,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00922,
+      "step": 72,
+      "tokens/total": 2179200,
+      "tokens/train_per_sec_per_gpu": 34.98,
+      "tokens/trainable": 32878
+    },
+    {
+      "epoch": 0.28515625,
+      "grad_norm": 0.7223671674728394,
+      "learning_rate": 9.794746487429161e-05,
+      "loss": 0.028303124010562897,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.02871,
+      "step": 73,
+      "tokens/total": 2209712,
+      "tokens/train_per_sec_per_gpu": 36.21,
+      "tokens/trainable": 33348
+    },
+    {
+      "epoch": 0.2890625,
+      "grad_norm": 0.6146635413169861,
+      "learning_rate": 9.785990005437554e-05,
+      "loss": 0.010147360153496265,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.0102,
+      "step": 74,
+      "tokens/total": 2240240,
+      "tokens/train_per_sec_per_gpu": 32.86,
+      "tokens/trainable": 33815
+    },
+    {
+      "epoch": 0.29296875,
+      "grad_norm": 0.28880107402801514,
+      "learning_rate": 9.777055165907117e-05,
+      "loss": 0.012021517381072044,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01209,
+      "step": 75,
+      "tokens/total": 2270336,
+      "tokens/train_per_sec_per_gpu": 34.86,
+      "tokens/trainable": 34229
+    },
+    {
+      "epoch": 0.296875,
+      "grad_norm": 0.4516862630844116,
+      "learning_rate": 9.767942340652993e-05,
+      "loss": 0.023413028568029404,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.02369,
+      "step": 76,
+      "tokens/total": 2300784,
+      "tokens/train_per_sec_per_gpu": 36.53,
+      "tokens/trainable": 34718
+    },
+    {
+      "epoch": 0.30078125,
+      "grad_norm": 0.9154849648475647,
+      "learning_rate": 9.758651908897035e-05,
+      "loss": 0.052833881229162216,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.76,
+      "memory/max_allocated (GiB)": 33.76,
+      "ppl": 1.05425,
+      "step": 77,
+      "tokens/total": 2330880,
+      "tokens/train_per_sec_per_gpu": 32.14,
+      "tokens/trainable": 35146
+    },
+    {
+      "epoch": 0.3046875,
+      "grad_norm": 0.32223156094551086,
+      "learning_rate": 9.749184257252033e-05,
+      "loss": 0.009072870016098022,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.00911,
+      "step": 78,
+      "tokens/total": 2361392,
+      "tokens/train_per_sec_per_gpu": 35.29,
+      "tokens/trainable": 35610
+    },
+    {
+      "epoch": 0.30859375,
+      "grad_norm": 0.4267147183418274,
+      "learning_rate": 9.739539779705614e-05,
+      "loss": 0.024432381615042686,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02473,
+      "step": 79,
+      "tokens/total": 2391744,
+      "tokens/train_per_sec_per_gpu": 29.95,
+      "tokens/trainable": 36037
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 0.3152504563331604,
+      "learning_rate": 9.729718877603861e-05,
+      "loss": 0.015125786885619164,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01524,
+      "step": 80,
+      "tokens/total": 2422240,
+      "tokens/train_per_sec_per_gpu": 38.73,
+      "tokens/trainable": 36522
+    },
+    {
+      "epoch": 0.31640625,
+      "grad_norm": 0.6696128249168396,
+      "learning_rate": 9.719721959634592e-05,
+      "loss": 0.023452740162611008,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.02373,
+      "step": 81,
+      "tokens/total": 2452560,
+      "tokens/train_per_sec_per_gpu": 33.67,
+      "tokens/trainable": 36979
+    },
+    {
+      "epoch": 0.3203125,
+      "grad_norm": 0.3085023760795593,
+      "learning_rate": 9.709549441810375e-05,
+      "loss": 0.011503729969263077,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.44,
+      "memory/max_allocated (GiB)": 33.44,
+      "ppl": 1.01157,
+      "step": 82,
+      "tokens/total": 2481024,
+      "tokens/train_per_sec_per_gpu": 33.93,
+      "tokens/trainable": 37439
+    },
+    {
+      "epoch": 0.32421875,
+      "grad_norm": 0.44436803460121155,
+      "learning_rate": 9.699201747451195e-05,
+      "loss": 0.01643884740769863,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01657,
+      "step": 83,
+      "tokens/total": 2511296,
+      "tokens/train_per_sec_per_gpu": 33.54,
+      "tokens/trainable": 37892
+    },
+    {
+      "epoch": 0.328125,
+      "grad_norm": 1.1241978406906128,
+      "learning_rate": 9.688679307166854e-05,
+      "loss": 0.0325784869492054,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.03311,
+      "step": 84,
+      "tokens/total": 2541792,
+      "tokens/train_per_sec_per_gpu": 33.19,
+      "tokens/trainable": 38355
+    },
+    {
+      "epoch": 0.33203125,
+      "grad_norm": 0.493679016828537,
+      "learning_rate": 9.677982558839042e-05,
+      "loss": 0.0119265615940094,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.69,
+      "memory/max_allocated (GiB)": 33.69,
+      "ppl": 1.012,
+      "step": 85,
+      "tokens/total": 2572032,
+      "tokens/train_per_sec_per_gpu": 34.07,
+      "tokens/trainable": 38810
+    },
+    {
+      "epoch": 0.3359375,
+      "grad_norm": 1.3241630792617798,
+      "learning_rate": 9.66711194760312e-05,
+      "loss": 0.04201122000813484,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.04291,
+      "step": 86,
+      "tokens/total": 2602512,
+      "tokens/train_per_sec_per_gpu": 30.89,
+      "tokens/trainable": 39240
+    },
+    {
+      "epoch": 0.33984375,
+      "grad_norm": 0.4352894723415375,
+      "learning_rate": 9.656067925829593e-05,
+      "loss": 0.028608884662389755,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.02902,
+      "step": 87,
+      "tokens/total": 2632880,
+      "tokens/train_per_sec_per_gpu": 37.0,
+      "tokens/trainable": 39748
+    },
+    {
+      "epoch": 0.34375,
+      "grad_norm": 0.48992764949798584,
+      "learning_rate": 9.644850953105288e-05,
+      "loss": 0.035998307168483734,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.03665,
+      "step": 88,
+      "tokens/total": 2663040,
+      "tokens/train_per_sec_per_gpu": 33.35,
+      "tokens/trainable": 40188
+    },
+    {
+      "epoch": 0.34765625,
+      "grad_norm": 0.491361141204834,
+      "learning_rate": 9.633461496214225e-05,
+      "loss": 0.02381829358637333,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0241,
+      "step": 89,
+      "tokens/total": 2693504,
+      "tokens/train_per_sec_per_gpu": 33.36,
+      "tokens/trainable": 40632
+    },
+    {
+      "epoch": 0.3515625,
+      "grad_norm": 0.4422471225261688,
+      "learning_rate": 9.621900029118195e-05,
+      "loss": 0.022936182096600533,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.0232,
+      "step": 90,
+      "tokens/total": 2723792,
+      "tokens/train_per_sec_per_gpu": 31.37,
+      "tokens/trainable": 41050
+    },
+    {
+      "epoch": 0.35546875,
+      "grad_norm": 0.4762331247329712,
+      "learning_rate": 9.610167032937036e-05,
+      "loss": 0.017587462440133095,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01774,
+      "step": 91,
+      "tokens/total": 2754048,
+      "tokens/train_per_sec_per_gpu": 37.05,
+      "tokens/trainable": 41515
+    },
+    {
+      "epoch": 0.359375,
+      "grad_norm": 0.4137004017829895,
+      "learning_rate": 9.598262995928611e-05,
+      "loss": 0.03153759241104126,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.03204,
+      "step": 92,
+      "tokens/total": 2784480,
+      "tokens/train_per_sec_per_gpu": 36.23,
+      "tokens/trainable": 41972
+    },
+    {
+      "epoch": 0.36328125,
+      "grad_norm": 0.5270156264305115,
+      "learning_rate": 9.586188413468492e-05,
+      "loss": 0.023751404136419296,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02404,
+      "step": 93,
+      "tokens/total": 2814928,
+      "tokens/train_per_sec_per_gpu": 35.43,
+      "tokens/trainable": 42442
+    },
+    {
+      "epoch": 0.3671875,
+      "grad_norm": 0.30221623182296753,
+      "learning_rate": 9.57394378802934e-05,
+      "loss": 0.012607569806277752,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01269,
+      "step": 94,
+      "tokens/total": 2845408,
+      "tokens/train_per_sec_per_gpu": 37.18,
+      "tokens/trainable": 42928
+    },
+    {
+      "epoch": 0.37109375,
+      "grad_norm": 0.34380587935447693,
+      "learning_rate": 9.56152962916e-05,
+      "loss": 0.021355444565415382,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02159,
+      "step": 95,
+      "tokens/total": 2875760,
+      "tokens/train_per_sec_per_gpu": 39.28,
+      "tokens/trainable": 43426
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 0.5729417204856873,
+      "learning_rate": 9.548946453464296e-05,
+      "loss": 0.011794034391641617,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01186,
+      "step": 96,
+      "tokens/total": 2906096,
+      "tokens/train_per_sec_per_gpu": 33.52,
+      "tokens/trainable": 43869
+    },
+    {
+      "epoch": 0.37890625,
+      "grad_norm": 0.1258164346218109,
+      "learning_rate": 9.53619478457953e-05,
+      "loss": 0.002208392135798931,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00221,
+      "step": 97,
+      "tokens/total": 2936560,
+      "tokens/train_per_sec_per_gpu": 32.02,
+      "tokens/trainable": 44300
+    },
+    {
+      "epoch": 0.3828125,
+      "grad_norm": 0.43118366599082947,
+      "learning_rate": 9.523275153154695e-05,
+      "loss": 0.015600357204675674,
+      "memory/device_reserved (GiB)": 35.19,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01572,
+      "step": 98,
+      "tokens/total": 2966880,
+      "tokens/train_per_sec_per_gpu": 39.37,
+      "tokens/trainable": 44791
+    },
+    {
+      "epoch": 0.38671875,
+      "grad_norm": 0.505718469619751,
+      "learning_rate": 9.51018809682839e-05,
+      "loss": 0.011874521151185036,
+      "memory/device_reserved (GiB)": 35.76,
+      "memory/max_active (GiB)": 34.03,
+      "memory/max_allocated (GiB)": 34.03,
+      "ppl": 1.01195,
+      "step": 99,
+      "tokens/total": 2997472,
+      "tokens/train_per_sec_per_gpu": 36.71,
+      "tokens/trainable": 45254
+    },
+    {
+      "epoch": 0.390625,
+      "grad_norm": 3.060197591781616,
+      "learning_rate": 9.49693416020645e-05,
+      "loss": 0.014575858600437641,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01468,
+      "step": 100,
+      "tokens/total": 3027840,
+      "tokens/train_per_sec_per_gpu": 33.57,
+      "tokens/trainable": 45698
+    },
+    {
+      "epoch": 0.39453125,
+      "grad_norm": 0.38254642486572266,
+      "learning_rate": 9.483513894839276e-05,
+      "loss": 0.009431221522390842,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00948,
+      "step": 101,
+      "tokens/total": 3058080,
+      "tokens/train_per_sec_per_gpu": 33.57,
+      "tokens/trainable": 46141
+    },
+    {
+      "epoch": 0.3984375,
+      "grad_norm": 0.8782851099967957,
+      "learning_rate": 9.469927859198888e-05,
+      "loss": 0.05128197371959686,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.05262,
+      "step": 102,
+      "tokens/total": 3088656,
+      "tokens/train_per_sec_per_gpu": 33.05,
+      "tokens/trainable": 46598
+    },
+    {
+      "epoch": 0.40234375,
+      "grad_norm": 0.42221176624298096,
+      "learning_rate": 9.456176618655689e-05,
+      "loss": 0.009157870896160603,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.0092,
+      "step": 103,
+      "tokens/total": 3119232,
+      "tokens/train_per_sec_per_gpu": 32.45,
+      "tokens/trainable": 47065
+    },
+    {
+      "epoch": 0.40625,
+      "grad_norm": 1.2625210285186768,
+      "learning_rate": 9.442260745454927e-05,
+      "loss": 0.01384878158569336,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01395,
+      "step": 104,
+      "tokens/total": 3149504,
+      "tokens/train_per_sec_per_gpu": 38.29,
+      "tokens/trainable": 47554
+    },
+    {
+      "epoch": 0.41015625,
+      "grad_norm": 0.3072832226753235,
+      "learning_rate": 9.428180818692884e-05,
+      "loss": 0.00823851116001606,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00827,
+      "step": 105,
+      "tokens/total": 3179872,
+      "tokens/train_per_sec_per_gpu": 36.88,
+      "tokens/trainable": 48047
+    },
+    {
+      "epoch": 0.4140625,
+      "grad_norm": 0.7415916323661804,
+      "learning_rate": 9.413937424292791e-05,
+      "loss": 0.03210819885134697,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.03263,
+      "step": 106,
+      "tokens/total": 3208128,
+      "tokens/train_per_sec_per_gpu": 34.62,
+      "tokens/trainable": 48498
+    },
+    {
+      "epoch": 0.41796875,
+      "grad_norm": 0.25024500489234924,
+      "learning_rate": 9.399531154980424e-05,
+      "loss": 0.00918180774897337,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.00922,
+      "step": 107,
+      "tokens/total": 3238224,
+      "tokens/train_per_sec_per_gpu": 31.18,
+      "tokens/trainable": 48933
+    },
+    {
+      "epoch": 0.421875,
+      "grad_norm": 0.5457918047904968,
+      "learning_rate": 9.384962610259455e-05,
+      "loss": 0.015052877366542816,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01517,
+      "step": 108,
+      "tokens/total": 3268592,
+      "tokens/train_per_sec_per_gpu": 38.48,
+      "tokens/trainable": 49404
+    },
+    {
+      "epoch": 0.42578125,
+      "grad_norm": 0.2508637607097626,
+      "learning_rate": 9.370232396386494e-05,
+      "loss": 0.007351120002567768,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00738,
+      "step": 109,
+      "tokens/total": 3298720,
+      "tokens/train_per_sec_per_gpu": 37.32,
+      "tokens/trainable": 49871
+    },
+    {
+      "epoch": 0.4296875,
+      "grad_norm": 0.2964157462120056,
+      "learning_rate": 9.355341126345868e-05,
+      "loss": 0.014206478372216225,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.01431,
+      "step": 110,
+      "tokens/total": 3329040,
+      "tokens/train_per_sec_per_gpu": 33.58,
+      "tokens/trainable": 50311
+    },
+    {
+      "epoch": 0.43359375,
+      "grad_norm": 0.3096546232700348,
+      "learning_rate": 9.340289419824107e-05,
+      "loss": 0.010700306855142117,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.01076,
+      "step": 111,
+      "tokens/total": 3359248,
+      "tokens/train_per_sec_per_gpu": 33.86,
+      "tokens/trainable": 50754
+    },
+    {
+      "epoch": 0.4375,
+      "grad_norm": 0.4973372220993042,
+      "learning_rate": 9.325077903184159e-05,
+      "loss": 0.013185497373342514,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01327,
+      "step": 112,
+      "tokens/total": 3389280,
+      "tokens/train_per_sec_per_gpu": 32.1,
+      "tokens/trainable": 51200
+    },
+    {
+      "epoch": 0.44140625,
+      "grad_norm": 0.4105257987976074,
+      "learning_rate": 9.30970720943932e-05,
+      "loss": 0.0069708251394331455,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.007,
+      "step": 113,
+      "tokens/total": 3419728,
+      "tokens/train_per_sec_per_gpu": 33.22,
+      "tokens/trainable": 51656
+    },
+    {
+      "epoch": 0.4453125,
+      "grad_norm": 0.2707289159297943,
+      "learning_rate": 9.2941779782269e-05,
+      "loss": 0.0033873419743031263,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.00339,
+      "step": 114,
+      "tokens/total": 3449984,
+      "tokens/train_per_sec_per_gpu": 30.92,
+      "tokens/trainable": 52069
+    },
+    {
+      "epoch": 0.44921875,
+      "grad_norm": 0.5602349042892456,
+      "learning_rate": 9.278490855781596e-05,
+      "loss": 0.00848664902150631,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00852,
+      "step": 115,
+      "tokens/total": 3480352,
+      "tokens/train_per_sec_per_gpu": 36.94,
+      "tokens/trainable": 52559
+    },
+    {
+      "epoch": 0.453125,
+      "grad_norm": 0.5224027037620544,
+      "learning_rate": 9.262646494908604e-05,
+      "loss": 0.02242768369615078,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02268,
+      "step": 116,
+      "tokens/total": 3510656,
+      "tokens/train_per_sec_per_gpu": 39.22,
+      "tokens/trainable": 53061
+    },
+    {
+      "epoch": 0.45703125,
+      "grad_norm": 1.2213423252105713,
+      "learning_rate": 9.246645554956457e-05,
+      "loss": 0.020464815199375153,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.02068,
+      "step": 117,
+      "tokens/total": 3541104,
+      "tokens/train_per_sec_per_gpu": 35.89,
+      "tokens/trainable": 53520
+    },
+    {
+      "epoch": 0.4609375,
+      "grad_norm": 1.30115807056427,
+      "learning_rate": 9.230488701789578e-05,
+      "loss": 0.025597743690013885,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.02593,
+      "step": 118,
+      "tokens/total": 3571744,
+      "tokens/train_per_sec_per_gpu": 33.36,
+      "tokens/trainable": 53976
+    },
+    {
+      "epoch": 0.46484375,
+      "grad_norm": 1.0988223552703857,
+      "learning_rate": 9.214176607760577e-05,
+      "loss": 0.02888210117816925,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.0293,
+      "step": 119,
+      "tokens/total": 3601808,
+      "tokens/train_per_sec_per_gpu": 34.93,
+      "tokens/trainable": 54408
+    },
+    {
+      "epoch": 0.46875,
+      "grad_norm": 0.3977462649345398,
+      "learning_rate": 9.197709951682268e-05,
+      "loss": 0.01678757183253765,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01693,
+      "step": 120,
+      "tokens/total": 3632096,
+      "tokens/train_per_sec_per_gpu": 37.14,
+      "tokens/trainable": 54884
+    },
+    {
+      "epoch": 0.47265625,
+      "grad_norm": 0.4872536063194275,
+      "learning_rate": 9.181089418799428e-05,
+      "loss": 0.022638533264398575,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.0229,
+      "step": 121,
+      "tokens/total": 3662416,
+      "tokens/train_per_sec_per_gpu": 33.15,
+      "tokens/trainable": 55303
+    },
+    {
+      "epoch": 0.4765625,
+      "grad_norm": 0.4663882255554199,
+      "learning_rate": 9.164315700760271e-05,
+      "loss": 0.023266036063432693,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.02354,
+      "step": 122,
+      "tokens/total": 3692992,
+      "tokens/train_per_sec_per_gpu": 31.94,
+      "tokens/trainable": 55743
+    },
+    {
+      "epoch": 0.48046875,
+      "grad_norm": 0.230746328830719,
+      "learning_rate": 9.147389495587671e-05,
+      "loss": 0.007810275070369244,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.00784,
+      "step": 123,
+      "tokens/total": 3722992,
+      "tokens/train_per_sec_per_gpu": 33.95,
+      "tokens/trainable": 56187
+    },
+    {
+      "epoch": 0.484375,
+      "grad_norm": 0.17318949103355408,
+      "learning_rate": 9.130311507650116e-05,
+      "loss": 0.006973203271627426,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.007,
+      "step": 124,
+      "tokens/total": 3753184,
+      "tokens/train_per_sec_per_gpu": 33.63,
+      "tokens/trainable": 56632
+    },
+    {
+      "epoch": 0.48828125,
+      "grad_norm": 0.19994892179965973,
+      "learning_rate": 9.113082447632394e-05,
+      "loss": 0.0068689389154314995,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00689,
+      "step": 125,
+      "tokens/total": 3783440,
+      "tokens/train_per_sec_per_gpu": 37.26,
+      "tokens/trainable": 57124
+    },
+    {
+      "epoch": 0.4921875,
+      "grad_norm": 0.11329996585845947,
+      "learning_rate": 9.09570303250602e-05,
+      "loss": 0.004266452044248581,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00428,
+      "step": 126,
+      "tokens/total": 3813840,
+      "tokens/train_per_sec_per_gpu": 33.38,
+      "tokens/trainable": 57570
+    },
+    {
+      "epoch": 0.49609375,
+      "grad_norm": 0.4156385362148285,
+      "learning_rate": 9.078173985499394e-05,
+      "loss": 0.02041183039546013,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02062,
+      "step": 127,
+      "tokens/total": 3844464,
+      "tokens/train_per_sec_per_gpu": 35.36,
+      "tokens/trainable": 58048
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.09819997847080231,
+      "learning_rate": 9.060496036067713e-05,
+      "loss": 0.0031540761701762676,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.00316,
+      "step": 128,
+      "tokens/total": 3872816,
+      "tokens/train_per_sec_per_gpu": 35.79,
+      "tokens/trainable": 58497
+    }
+  ],
+  "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": 2.628706536652631e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

aft_wave_v2/control_matched__charter2/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/control_matched__charter2/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/control_matched__charter2/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": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "gate_proj",
+    "v_proj",
+    "up_proj"
+  ],
+  "target_parameters": [],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-160/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb09b512ed97e9c1471f4bf12ff053743a9ae143b3eb48010e034fb0e01b2206
+size 547777976

aft_wave_v2/control_matched__charter2/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/control_matched__charter2/training/checkpoints/checkpoint-160/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53e9ba68f36679e6cfe39be8a9c321dc9071aa233fc3654a1f7aaf1ea9e50499
+size 1048106435

aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-160/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d3d8d99873a308b8540f12cd8c43a40b60230c9c2eccf247652e1b5b269679cd
+size 14645

aft_wave_v2/control_matched__charter2/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/control_matched__charter2/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/control_matched__charter2/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/control_matched__charter2/training/checkpoints/checkpoint-160/tokens_state.json

@@ -0,0 +1 @@
+{"total": 4836688, "trainable": 73119}

aft_wave_v2/control_matched__charter2/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": 8.088534355163574,
+      "learning_rate": 0.0,
+      "loss": 0.1503293514251709,
+      "memory/device_reserved (GiB)": 33.9,
+      "memory/max_active (GiB)": 32.79,
+      "memory/max_allocated (GiB)": 32.79,
+      "ppl": 1.16222,
+      "step": 1,
+      "tokens/total": 30176,
+      "tokens/train_per_sec_per_gpu": 28.32,
+      "tokens/trainable": 425
+    },
+    {
+      "epoch": 0.0078125,
+      "grad_norm": 0.8953529596328735,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.15210241079330444,
+      "memory/device_reserved (GiB)": 34.68,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.16428,
+      "step": 2,
+      "tokens/total": 60272,
+      "tokens/train_per_sec_per_gpu": 33.91,
+      "tokens/trainable": 872
+    },
+    {
+      "epoch": 0.01171875,
+      "grad_norm": 1.4013757705688477,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.14530743658542633,
+      "memory/device_reserved (GiB)": 34.72,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.1564,
+      "step": 3,
+      "tokens/total": 90512,
+      "tokens/train_per_sec_per_gpu": 35.5,
+      "tokens/trainable": 1352
+    },
+    {
+      "epoch": 0.015625,
+      "grad_norm": 0.9737329483032227,
+      "learning_rate": 1.2e-05,
+      "loss": 0.13504940271377563,
+      "memory/device_reserved (GiB)": 34.73,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.14459,
+      "step": 4,
+      "tokens/total": 120944,
+      "tokens/train_per_sec_per_gpu": 31.75,
+      "tokens/trainable": 1777
+    },
+    {
+      "epoch": 0.01953125,
+      "grad_norm": 1.5541704893112183,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.13814105093479156,
+      "memory/device_reserved (GiB)": 35.12,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.14814,
+      "step": 5,
+      "tokens/total": 151440,
+      "tokens/train_per_sec_per_gpu": 37.35,
+      "tokens/trainable": 2260
+    },
+    {
+      "epoch": 0.0234375,
+      "grad_norm": 6.433568477630615,
+      "learning_rate": 2e-05,
+      "loss": 0.12622183561325073,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.13453,
+      "step": 6,
+      "tokens/total": 181984,
+      "tokens/train_per_sec_per_gpu": 33.21,
+      "tokens/trainable": 2704
+    },
+    {
+      "epoch": 0.02734375,
+      "grad_norm": 1.4011743068695068,
+      "learning_rate": 2.4e-05,
+      "loss": 0.11199458688497543,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.11851,
+      "step": 7,
+      "tokens/total": 212336,
+      "tokens/train_per_sec_per_gpu": 34.37,
+      "tokens/trainable": 3143
+    },
+    {
+      "epoch": 0.03125,
+      "grad_norm": 2.038933038711548,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.09219667315483093,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.09658,
+      "step": 8,
+      "tokens/total": 242592,
+      "tokens/train_per_sec_per_gpu": 34.78,
+      "tokens/trainable": 3586
+    },
+    {
+      "epoch": 0.03515625,
+      "grad_norm": 4.82806396484375,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.04823308065533638,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.04942,
+      "step": 9,
+      "tokens/total": 272784,
+      "tokens/train_per_sec_per_gpu": 29.59,
+      "tokens/trainable": 4007
+    },
+    {
+      "epoch": 0.0390625,
+      "grad_norm": 1.3691715002059937,
+      "learning_rate": 3.6e-05,
+      "loss": 0.04049058258533478,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.04132,
+      "step": 10,
+      "tokens/total": 303184,
+      "tokens/train_per_sec_per_gpu": 39.66,
+      "tokens/trainable": 4481
+    },
+    {
+      "epoch": 0.04296875,
+      "grad_norm": 2.285440444946289,
+      "learning_rate": 4e-05,
+      "loss": 0.06339800357818604,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.06545,
+      "step": 11,
+      "tokens/total": 333296,
+      "tokens/train_per_sec_per_gpu": 35.09,
+      "tokens/trainable": 4958
+    },
+    {
+      "epoch": 0.046875,
+      "grad_norm": 5.0530476570129395,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.07100467383861542,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.07359,
+      "step": 12,
+      "tokens/total": 363840,
+      "tokens/train_per_sec_per_gpu": 34.81,
+      "tokens/trainable": 5440
+    },
+    {
+      "epoch": 0.05078125,
+      "grad_norm": 3.5007340908050537,
+      "learning_rate": 4.8e-05,
+      "loss": 0.061270229518413544,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.06319,
+      "step": 13,
+      "tokens/total": 394112,
+      "tokens/train_per_sec_per_gpu": 35.82,
+      "tokens/trainable": 5871
+    },
+    {
+      "epoch": 0.0546875,
+      "grad_norm": 2.6390483379364014,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.043673258274793625,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.04464,
+      "step": 14,
+      "tokens/total": 424656,
+      "tokens/train_per_sec_per_gpu": 35.29,
+      "tokens/trainable": 6360
+    },
+    {
+      "epoch": 0.05859375,
+      "grad_norm": 2.6278514862060547,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 0.06025860086083412,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.06211,
+      "step": 15,
+      "tokens/total": 455152,
+      "tokens/train_per_sec_per_gpu": 36.67,
+      "tokens/trainable": 6841
+    },
+    {
+      "epoch": 0.0625,
+      "grad_norm": 5.878053665161133,
+      "learning_rate": 6e-05,
+      "loss": 0.09122475981712341,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.09552,
+      "step": 16,
+      "tokens/total": 485328,
+      "tokens/train_per_sec_per_gpu": 31.7,
+      "tokens/trainable": 7254
+    },
+    {
+      "epoch": 0.06640625,
+      "grad_norm": 1.9717961549758911,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.017721600830554962,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01788,
+      "step": 17,
+      "tokens/total": 515744,
+      "tokens/train_per_sec_per_gpu": 36.52,
+      "tokens/trainable": 7723
+    },
+    {
+      "epoch": 0.0703125,
+      "grad_norm": 1.783061146736145,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.04740295931696892,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.04854,
+      "step": 18,
+      "tokens/total": 546144,
+      "tokens/train_per_sec_per_gpu": 34.38,
+      "tokens/trainable": 8208
+    },
+    {
+      "epoch": 0.07421875,
+      "grad_norm": 1.916291356086731,
+      "learning_rate": 7.2e-05,
+      "loss": 0.025028439238667488,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02534,
+      "step": 19,
+      "tokens/total": 576560,
+      "tokens/train_per_sec_per_gpu": 35.49,
+      "tokens/trainable": 8699
+    },
+    {
+      "epoch": 0.078125,
+      "grad_norm": 2.9791574478149414,
+      "learning_rate": 7.6e-05,
+      "loss": 0.055660396814346313,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.05724,
+      "step": 20,
+      "tokens/total": 607056,
+      "tokens/train_per_sec_per_gpu": 40.56,
+      "tokens/trainable": 9214
+    },
+    {
+      "epoch": 0.08203125,
+      "grad_norm": 1.2719122171401978,
+      "learning_rate": 8e-05,
+      "loss": 0.024366460740566254,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02467,
+      "step": 21,
+      "tokens/total": 637488,
+      "tokens/train_per_sec_per_gpu": 32.12,
+      "tokens/trainable": 9639
+    },
+    {
+      "epoch": 0.0859375,
+      "grad_norm": 2.241725444793701,
+      "learning_rate": 8.4e-05,
+      "loss": 0.05170319974422455,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.05306,
+      "step": 22,
+      "tokens/total": 667632,
+      "tokens/train_per_sec_per_gpu": 36.71,
+      "tokens/trainable": 10111
+    },
+    {
+      "epoch": 0.08984375,
+      "grad_norm": 2.2158987522125244,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.04347127676010132,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.04443,
+      "step": 23,
+      "tokens/total": 697872,
+      "tokens/train_per_sec_per_gpu": 33.4,
+      "tokens/trainable": 10560
+    },
+    {
+      "epoch": 0.09375,
+      "grad_norm": 1.028172254562378,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 0.016212783753871918,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01634,
+      "step": 24,
+      "tokens/total": 728432,
+      "tokens/train_per_sec_per_gpu": 32.0,
+      "tokens/trainable": 11018
+    },
+    {
+      "epoch": 0.09765625,
+      "grad_norm": 1.0408563613891602,
+      "learning_rate": 9.6e-05,
+      "loss": 0.023505806922912598,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02378,
+      "step": 25,
+      "tokens/total": 756656,
+      "tokens/train_per_sec_per_gpu": 40.53,
+      "tokens/trainable": 11442
+    },
+    {
+      "epoch": 0.1015625,
+      "grad_norm": 1.6656584739685059,
+      "learning_rate": 0.0001,
+      "loss": 0.05568547546863556,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.05727,
+      "step": 26,
+      "tokens/total": 787120,
+      "tokens/train_per_sec_per_gpu": 34.83,
+      "tokens/trainable": 11899
+    },
+    {
+      "epoch": 0.10546875,
+      "grad_norm": 0.7106256484985352,
+      "learning_rate": 9.99990636831587e-05,
+      "loss": 0.009144511073827744,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.00919,
+      "step": 27,
+      "tokens/total": 817504,
+      "tokens/train_per_sec_per_gpu": 32.1,
+      "tokens/trainable": 12341
+    },
+    {
+      "epoch": 0.109375,
+      "grad_norm": 0.8463047742843628,
+      "learning_rate": 9.999625477159879e-05,
+      "loss": 0.020250165835022926,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02046,
+      "step": 28,
+      "tokens/total": 848128,
+      "tokens/train_per_sec_per_gpu": 38.39,
+      "tokens/trainable": 12825
+    },
+    {
+      "epoch": 0.11328125,
+      "grad_norm": 1.0744967460632324,
+      "learning_rate": 9.999157338221051e-05,
+      "loss": 0.023447610437870026,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02372,
+      "step": 29,
+      "tokens/total": 878448,
+      "tokens/train_per_sec_per_gpu": 36.01,
+      "tokens/trainable": 13295
+    },
+    {
+      "epoch": 0.1171875,
+      "grad_norm": 1.4234495162963867,
+      "learning_rate": 9.998501970980562e-05,
+      "loss": 0.03458714857697487,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.03519,
+      "step": 30,
+      "tokens/total": 909008,
+      "tokens/train_per_sec_per_gpu": 35.8,
+      "tokens/trainable": 13784
+    },
+    {
+      "epoch": 0.12109375,
+      "grad_norm": 0.9880445003509521,
+      "learning_rate": 9.997659402710915e-05,
+      "loss": 0.02479860559105873,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02511,
+      "step": 31,
+      "tokens/total": 939312,
+      "tokens/train_per_sec_per_gpu": 34.58,
+      "tokens/trainable": 14250
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 1.1164203882217407,
+      "learning_rate": 9.996629668474818e-05,
+      "loss": 0.023022079840302467,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02329,
+      "step": 32,
+      "tokens/total": 969264,
+      "tokens/train_per_sec_per_gpu": 35.0,
+      "tokens/trainable": 14680
+    },
+    {
+      "epoch": 0.12890625,
+      "grad_norm": 1.324739694595337,
+      "learning_rate": 9.995412811123711e-05,
+      "loss": 0.0630958303809166,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.06513,
+      "step": 33,
+      "tokens/total": 999792,
+      "tokens/train_per_sec_per_gpu": 33.83,
+      "tokens/trainable": 15112
+    },
+    {
+      "epoch": 0.1328125,
+      "grad_norm": 1.0034446716308594,
+      "learning_rate": 9.994008881295999e-05,
+      "loss": 0.04252947121858597,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.04345,
+      "step": 34,
+      "tokens/total": 1029840,
+      "tokens/train_per_sec_per_gpu": 34.46,
+      "tokens/trainable": 15573
+    },
+    {
+      "epoch": 0.13671875,
+      "grad_norm": 1.1391361951828003,
+      "learning_rate": 9.992417937414932e-05,
+      "loss": 0.0361175499856472,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03678,
+      "step": 35,
+      "tokens/total": 1060416,
+      "tokens/train_per_sec_per_gpu": 34.24,
+      "tokens/trainable": 16062
+    },
+    {
+      "epoch": 0.140625,
+      "grad_norm": 0.48655080795288086,
+      "learning_rate": 9.99064004568618e-05,
+      "loss": 0.015174772590398788,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.72,
+      "memory/max_allocated (GiB)": 33.72,
+      "ppl": 1.01529,
+      "step": 36,
+      "tokens/total": 1090464,
+      "tokens/train_per_sec_per_gpu": 31.53,
+      "tokens/trainable": 16500
+    },
+    {
+      "epoch": 0.14453125,
+      "grad_norm": 0.7631387114524841,
+      "learning_rate": 9.988675280095074e-05,
+      "loss": 0.04621092230081558,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.0473,
+      "step": 37,
+      "tokens/total": 1120944,
+      "tokens/train_per_sec_per_gpu": 34.9,
+      "tokens/trainable": 16936
+    },
+    {
+      "epoch": 0.1484375,
+      "grad_norm": 1.1162818670272827,
+      "learning_rate": 9.986523722403528e-05,
+      "loss": 0.046002037823200226,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.04708,
+      "step": 38,
+      "tokens/total": 1151360,
+      "tokens/train_per_sec_per_gpu": 29.32,
+      "tokens/trainable": 17357
+    },
+    {
+      "epoch": 0.15234375,
+      "grad_norm": 4.360172271728516,
+      "learning_rate": 9.984185462146642e-05,
+      "loss": 0.052004992961883545,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.05338,
+      "step": 39,
+      "tokens/total": 1181728,
+      "tokens/train_per_sec_per_gpu": 36.82,
+      "tokens/trainable": 17831
+    },
+    {
+      "epoch": 0.15625,
+      "grad_norm": 0.8433209657669067,
+      "learning_rate": 9.98166059662897e-05,
+      "loss": 0.04868919029831886,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.04989,
+      "step": 40,
+      "tokens/total": 1212064,
+      "tokens/train_per_sec_per_gpu": 36.15,
+      "tokens/trainable": 18315
+    },
+    {
+      "epoch": 0.16015625,
+      "grad_norm": 0.9528040289878845,
+      "learning_rate": 9.978949230920472e-05,
+      "loss": 0.029145417734980583,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.02957,
+      "step": 41,
+      "tokens/total": 1242592,
+      "tokens/train_per_sec_per_gpu": 32.49,
+      "tokens/trainable": 18772
+    },
+    {
+      "epoch": 0.1640625,
+      "grad_norm": 0.5701055526733398,
+      "learning_rate": 9.976051477852141e-05,
+      "loss": 0.01586967147886753,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.016,
+      "step": 42,
+      "tokens/total": 1272800,
+      "tokens/train_per_sec_per_gpu": 32.32,
+      "tokens/trainable": 19197
+    },
+    {
+      "epoch": 0.16796875,
+      "grad_norm": 0.372957706451416,
+      "learning_rate": 9.972967458011312e-05,
+      "loss": 0.01052568107843399,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 34.04,
+      "memory/max_allocated (GiB)": 34.04,
+      "ppl": 1.01058,
+      "step": 43,
+      "tokens/total": 1303600,
+      "tokens/train_per_sec_per_gpu": 31.68,
+      "tokens/trainable": 19647
+    },
+    {
+      "epoch": 0.171875,
+      "grad_norm": 0.550298810005188,
+      "learning_rate": 9.96969729973664e-05,
+      "loss": 0.019961029291152954,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02016,
+      "step": 44,
+      "tokens/total": 1333936,
+      "tokens/train_per_sec_per_gpu": 33.83,
+      "tokens/trainable": 20097
+    },
+    {
+      "epoch": 0.17578125,
+      "grad_norm": 3.7621572017669678,
+      "learning_rate": 9.966241139112754e-05,
+      "loss": 0.06465393304824829,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.06679,
+      "step": 45,
+      "tokens/total": 1364288,
+      "tokens/train_per_sec_per_gpu": 31.92,
+      "tokens/trainable": 20529
+    },
+    {
+      "epoch": 0.1796875,
+      "grad_norm": 1.1490390300750732,
+      "learning_rate": 9.96259911996461e-05,
+      "loss": 0.055192310363054276,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.05674,
+      "step": 46,
+      "tokens/total": 1394432,
+      "tokens/train_per_sec_per_gpu": 34.68,
+      "tokens/trainable": 20973
+    },
+    {
+      "epoch": 0.18359375,
+      "grad_norm": 0.5583573579788208,
+      "learning_rate": 9.958771393851491e-05,
+      "loss": 0.024018850177526474,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.98,
+      "memory/max_allocated (GiB)": 33.98,
+      "ppl": 1.02431,
+      "step": 47,
+      "tokens/total": 1425088,
+      "tokens/train_per_sec_per_gpu": 35.36,
+      "tokens/trainable": 21437
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 0.7726427912712097,
+      "learning_rate": 9.954758120060702e-05,
+      "loss": 0.021447816863656044,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.02168,
+      "step": 48,
+      "tokens/total": 1455456,
+      "tokens/train_per_sec_per_gpu": 35.57,
+      "tokens/trainable": 21860
+    },
+    {
+      "epoch": 0.19140625,
+      "grad_norm": 0.9507198929786682,
+      "learning_rate": 9.950559465600948e-05,
+      "loss": 0.03307538479566574,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.7,
+      "memory/max_allocated (GiB)": 33.7,
+      "ppl": 1.03363,
+      "step": 49,
+      "tokens/total": 1485376,
+      "tokens/train_per_sec_per_gpu": 30.1,
+      "tokens/trainable": 22273
+    },
+    {
+      "epoch": 0.1953125,
+      "grad_norm": 0.5201236009597778,
+      "learning_rate": 9.946175605195379e-05,
+      "loss": 0.019475962966680527,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01967,
+      "step": 50,
+      "tokens/total": 1515744,
+      "tokens/train_per_sec_per_gpu": 36.83,
+      "tokens/trainable": 22720
+    },
+    {
+      "epoch": 0.19921875,
+      "grad_norm": 0.5436747670173645,
+      "learning_rate": 9.941606721274322e-05,
+      "loss": 0.021235385909676552,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.02146,
+      "step": 51,
+      "tokens/total": 1545984,
+      "tokens/train_per_sec_per_gpu": 32.55,
+      "tokens/trainable": 23168
+    },
+    {
+      "epoch": 0.203125,
+      "grad_norm": 0.47054970264434814,
+      "learning_rate": 9.936853003967685e-05,
+      "loss": 0.02223706617951393,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.02249,
+      "step": 52,
+      "tokens/total": 1576288,
+      "tokens/train_per_sec_per_gpu": 37.44,
+      "tokens/trainable": 23663
+    },
+    {
+      "epoch": 0.20703125,
+      "grad_norm": 12.063824653625488,
+      "learning_rate": 9.93191465109705e-05,
+      "loss": 0.01991445943713188,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02011,
+      "step": 53,
+      "tokens/total": 1606528,
+      "tokens/train_per_sec_per_gpu": 34.49,
+      "tokens/trainable": 24128
+    },
+    {
+      "epoch": 0.2109375,
+      "grad_norm": 0.731513500213623,
+      "learning_rate": 9.926791868167438e-05,
+      "loss": 0.035772789269685745,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.66,
+      "memory/max_allocated (GiB)": 33.66,
+      "ppl": 1.03642,
+      "step": 54,
+      "tokens/total": 1636400,
+      "tokens/train_per_sec_per_gpu": 38.33,
+      "tokens/trainable": 24601
+    },
+    {
+      "epoch": 0.21484375,
+      "grad_norm": 0.8308107256889343,
+      "learning_rate": 9.921484868358753e-05,
+      "loss": 0.04387127608060837,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.04485,
+      "step": 55,
+      "tokens/total": 1666896,
+      "tokens/train_per_sec_per_gpu": 34.44,
+      "tokens/trainable": 25099
+    },
+    {
+      "epoch": 0.21875,
+      "grad_norm": 0.3609774708747864,
+      "learning_rate": 9.915993872516924e-05,
+      "loss": 0.009335450828075409,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00938,
+      "step": 56,
+      "tokens/total": 1697232,
+      "tokens/train_per_sec_per_gpu": 37.08,
+      "tokens/trainable": 25573
+    },
+    {
+      "epoch": 0.22265625,
+      "grad_norm": 0.5462154150009155,
+      "learning_rate": 9.9103191091447e-05,
+      "loss": 0.016265662387013435,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.0164,
+      "step": 57,
+      "tokens/total": 1727440,
+      "tokens/train_per_sec_per_gpu": 33.69,
+      "tokens/trainable": 25997
+    },
+    {
+      "epoch": 0.2265625,
+      "grad_norm": 3.2781224250793457,
+      "learning_rate": 9.904460814392147e-05,
+      "loss": 0.015918206423521042,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.01605,
+      "step": 58,
+      "tokens/total": 1757952,
+      "tokens/train_per_sec_per_gpu": 34.37,
+      "tokens/trainable": 26441
+    },
+    {
+      "epoch": 0.23046875,
+      "grad_norm": 1.6426713466644287,
+      "learning_rate": 9.898419232046825e-05,
+      "loss": 0.0164799727499485,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01662,
+      "step": 59,
+      "tokens/total": 1788576,
+      "tokens/train_per_sec_per_gpu": 34.54,
+      "tokens/trainable": 26915
+    },
+    {
+      "epoch": 0.234375,
+      "grad_norm": 0.5991567969322205,
+      "learning_rate": 9.892194613523633e-05,
+      "loss": 0.010309271514415741,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01036,
+      "step": 60,
+      "tokens/total": 1818816,
+      "tokens/train_per_sec_per_gpu": 34.2,
+      "tokens/trainable": 27355
+    },
+    {
+      "epoch": 0.23828125,
+      "grad_norm": 0.12077115476131439,
+      "learning_rate": 9.885787217854357e-05,
+      "loss": 0.00214880402199924,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00215,
+      "step": 61,
+      "tokens/total": 1849424,
+      "tokens/train_per_sec_per_gpu": 39.71,
+      "tokens/trainable": 27862
+    },
+    {
+      "epoch": 0.2421875,
+      "grad_norm": 0.8362685441970825,
+      "learning_rate": 9.879197311676887e-05,
+      "loss": 0.04497821629047394,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 34.02,
+      "memory/max_allocated (GiB)": 34.02,
+      "ppl": 1.04601,
+      "step": 62,
+      "tokens/total": 1880032,
+      "tokens/train_per_sec_per_gpu": 32.8,
+      "tokens/trainable": 28313
+    },
+    {
+      "epoch": 0.24609375,
+      "grad_norm": 1.2074859142303467,
+      "learning_rate": 9.872425169224113e-05,
+      "loss": 0.04683143272995949,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.04795,
+      "step": 63,
+      "tokens/total": 1910512,
+      "tokens/train_per_sec_per_gpu": 32.7,
+      "tokens/trainable": 28774
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8954252600669861,
+      "learning_rate": 9.865471072312528e-05,
+      "loss": 0.028329109773039818,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02873,
+      "step": 64,
+      "tokens/total": 1940608,
+      "tokens/train_per_sec_per_gpu": 34.07,
+      "tokens/trainable": 29222
+    },
+    {
+      "epoch": 0.25390625,
+      "grad_norm": 0.710774838924408,
+      "learning_rate": 9.858335310330492e-05,
+      "loss": 0.018383217975497246,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.01855,
+      "step": 65,
+      "tokens/total": 1971104,
+      "tokens/train_per_sec_per_gpu": 32.0,
+      "tokens/trainable": 29687
+    },
+    {
+      "epoch": 0.2578125,
+      "grad_norm": 0.841823935508728,
+      "learning_rate": 9.851018180226185e-05,
+      "loss": 0.024537263438105583,
+      "memory/device_reserved (GiB)": 35.06,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02484,
+      "step": 66,
+      "tokens/total": 2001472,
+      "tokens/train_per_sec_per_gpu": 33.93,
+      "tokens/trainable": 30114
+    },
+    {
+      "epoch": 0.26171875,
+      "grad_norm": 0.5852411389350891,
+      "learning_rate": 9.843519986495259e-05,
+      "loss": 0.03270117565989494,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.03324,
+      "step": 67,
+      "tokens/total": 2029696,
+      "tokens/train_per_sec_per_gpu": 38.04,
+      "tokens/trainable": 30594
+    },
+    {
+      "epoch": 0.265625,
+      "grad_norm": 0.3637949824333191,
+      "learning_rate": 9.835841041168162e-05,
+      "loss": 0.011452089995145798,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01152,
+      "step": 68,
+      "tokens/total": 2060144,
+      "tokens/train_per_sec_per_gpu": 35.62,
+      "tokens/trainable": 31073
+    },
+    {
+      "epoch": 0.26953125,
+      "grad_norm": 0.4834350049495697,
+      "learning_rate": 9.82798166379715e-05,
+      "loss": 0.032057274132966995,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.03258,
+      "step": 69,
+      "tokens/total": 2090448,
+      "tokens/train_per_sec_per_gpu": 34.38,
+      "tokens/trainable": 31553
+    },
+    {
+      "epoch": 0.2734375,
+      "grad_norm": 0.5799927115440369,
+      "learning_rate": 9.819942181443002e-05,
+      "loss": 0.016432201489806175,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01657,
+      "step": 70,
+      "tokens/total": 2120608,
+      "tokens/train_per_sec_per_gpu": 32.72,
+      "tokens/trainable": 32004
+    },
+    {
+      "epoch": 0.27734375,
+      "grad_norm": 0.6350996494293213,
+      "learning_rate": 9.811722928661392e-05,
+      "loss": 0.04276867210865021,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.35,
+      "memory/max_allocated (GiB)": 33.35,
+      "ppl": 1.0437,
+      "step": 71,
+      "tokens/total": 2148848,
+      "tokens/train_per_sec_per_gpu": 32.26,
+      "tokens/trainable": 32419
+    },
+    {
+      "epoch": 0.28125,
+      "grad_norm": 0.2791362702846527,
+      "learning_rate": 9.803324247488975e-05,
+      "loss": 0.009182551875710487,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00922,
+      "step": 72,
+      "tokens/total": 2179200,
+      "tokens/train_per_sec_per_gpu": 34.98,
+      "tokens/trainable": 32878
+    },
+    {
+      "epoch": 0.28515625,
+      "grad_norm": 0.7223671674728394,
+      "learning_rate": 9.794746487429161e-05,
+      "loss": 0.028303124010562897,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.02871,
+      "step": 73,
+      "tokens/total": 2209712,
+      "tokens/train_per_sec_per_gpu": 36.21,
+      "tokens/trainable": 33348
+    },
+    {
+      "epoch": 0.2890625,
+      "grad_norm": 0.6146635413169861,
+      "learning_rate": 9.785990005437554e-05,
+      "loss": 0.010147360153496265,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.0102,
+      "step": 74,
+      "tokens/total": 2240240,
+      "tokens/train_per_sec_per_gpu": 32.86,
+      "tokens/trainable": 33815
+    },
+    {
+      "epoch": 0.29296875,
+      "grad_norm": 0.28880107402801514,
+      "learning_rate": 9.777055165907117e-05,
+      "loss": 0.012021517381072044,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01209,
+      "step": 75,
+      "tokens/total": 2270336,
+      "tokens/train_per_sec_per_gpu": 34.86,
+      "tokens/trainable": 34229
+    },
+    {
+      "epoch": 0.296875,
+      "grad_norm": 0.4516862630844116,
+      "learning_rate": 9.767942340652993e-05,
+      "loss": 0.023413028568029404,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.02369,
+      "step": 76,
+      "tokens/total": 2300784,
+      "tokens/train_per_sec_per_gpu": 36.53,
+      "tokens/trainable": 34718
+    },
+    {
+      "epoch": 0.30078125,
+      "grad_norm": 0.9154849648475647,
+      "learning_rate": 9.758651908897035e-05,
+      "loss": 0.052833881229162216,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.76,
+      "memory/max_allocated (GiB)": 33.76,
+      "ppl": 1.05425,
+      "step": 77,
+      "tokens/total": 2330880,
+      "tokens/train_per_sec_per_gpu": 32.14,
+      "tokens/trainable": 35146
+    },
+    {
+      "epoch": 0.3046875,
+      "grad_norm": 0.32223156094551086,
+      "learning_rate": 9.749184257252033e-05,
+      "loss": 0.009072870016098022,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.00911,
+      "step": 78,
+      "tokens/total": 2361392,
+      "tokens/train_per_sec_per_gpu": 35.29,
+      "tokens/trainable": 35610
+    },
+    {
+      "epoch": 0.30859375,
+      "grad_norm": 0.4267147183418274,
+      "learning_rate": 9.739539779705614e-05,
+      "loss": 0.024432381615042686,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02473,
+      "step": 79,
+      "tokens/total": 2391744,
+      "tokens/train_per_sec_per_gpu": 29.95,
+      "tokens/trainable": 36037
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 0.3152504563331604,
+      "learning_rate": 9.729718877603861e-05,
+      "loss": 0.015125786885619164,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01524,
+      "step": 80,
+      "tokens/total": 2422240,
+      "tokens/train_per_sec_per_gpu": 38.73,
+      "tokens/trainable": 36522
+    },
+    {
+      "epoch": 0.31640625,
+      "grad_norm": 0.6696128249168396,
+      "learning_rate": 9.719721959634592e-05,
+      "loss": 0.023452740162611008,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.02373,
+      "step": 81,
+      "tokens/total": 2452560,
+      "tokens/train_per_sec_per_gpu": 33.67,
+      "tokens/trainable": 36979
+    },
+    {
+      "epoch": 0.3203125,
+      "grad_norm": 0.3085023760795593,
+      "learning_rate": 9.709549441810375e-05,
+      "loss": 0.011503729969263077,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.44,
+      "memory/max_allocated (GiB)": 33.44,
+      "ppl": 1.01157,
+      "step": 82,
+      "tokens/total": 2481024,
+      "tokens/train_per_sec_per_gpu": 33.93,
+      "tokens/trainable": 37439
+    },
+    {
+      "epoch": 0.32421875,
+      "grad_norm": 0.44436803460121155,
+      "learning_rate": 9.699201747451195e-05,
+      "loss": 0.01643884740769863,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01657,
+      "step": 83,
+      "tokens/total": 2511296,
+      "tokens/train_per_sec_per_gpu": 33.54,
+      "tokens/trainable": 37892
+    },
+    {
+      "epoch": 0.328125,
+      "grad_norm": 1.1241978406906128,
+      "learning_rate": 9.688679307166854e-05,
+      "loss": 0.0325784869492054,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.03311,
+      "step": 84,
+      "tokens/total": 2541792,
+      "tokens/train_per_sec_per_gpu": 33.19,
+      "tokens/trainable": 38355
+    },
+    {
+      "epoch": 0.33203125,
+      "grad_norm": 0.493679016828537,
+      "learning_rate": 9.677982558839042e-05,
+      "loss": 0.0119265615940094,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.69,
+      "memory/max_allocated (GiB)": 33.69,
+      "ppl": 1.012,
+      "step": 85,
+      "tokens/total": 2572032,
+      "tokens/train_per_sec_per_gpu": 34.07,
+      "tokens/trainable": 38810
+    },
+    {
+      "epoch": 0.3359375,
+      "grad_norm": 1.3241630792617798,
+      "learning_rate": 9.66711194760312e-05,
+      "loss": 0.04201122000813484,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.04291,
+      "step": 86,
+      "tokens/total": 2602512,
+      "tokens/train_per_sec_per_gpu": 30.89,
+      "tokens/trainable": 39240
+    },
+    {
+      "epoch": 0.33984375,
+      "grad_norm": 0.4352894723415375,
+      "learning_rate": 9.656067925829593e-05,
+      "loss": 0.028608884662389755,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.02902,
+      "step": 87,
+      "tokens/total": 2632880,
+      "tokens/train_per_sec_per_gpu": 37.0,
+      "tokens/trainable": 39748
+    },
+    {
+      "epoch": 0.34375,
+      "grad_norm": 0.48992764949798584,
+      "learning_rate": 9.644850953105288e-05,
+      "loss": 0.035998307168483734,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.03665,
+      "step": 88,
+      "tokens/total": 2663040,
+      "tokens/train_per_sec_per_gpu": 33.35,
+      "tokens/trainable": 40188
+    },
+    {
+      "epoch": 0.34765625,
+      "grad_norm": 0.491361141204834,
+      "learning_rate": 9.633461496214225e-05,
+      "loss": 0.02381829358637333,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0241,
+      "step": 89,
+      "tokens/total": 2693504,
+      "tokens/train_per_sec_per_gpu": 33.36,
+      "tokens/trainable": 40632
+    },
+    {
+      "epoch": 0.3515625,
+      "grad_norm": 0.4422471225261688,
+      "learning_rate": 9.621900029118195e-05,
+      "loss": 0.022936182096600533,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.0232,
+      "step": 90,
+      "tokens/total": 2723792,
+      "tokens/train_per_sec_per_gpu": 31.37,
+      "tokens/trainable": 41050
+    },
+    {
+      "epoch": 0.35546875,
+      "grad_norm": 0.4762331247329712,
+      "learning_rate": 9.610167032937036e-05,
+      "loss": 0.017587462440133095,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01774,
+      "step": 91,
+      "tokens/total": 2754048,
+      "tokens/train_per_sec_per_gpu": 37.05,
+      "tokens/trainable": 41515
+    },
+    {
+      "epoch": 0.359375,
+      "grad_norm": 0.4137004017829895,
+      "learning_rate": 9.598262995928611e-05,
+      "loss": 0.03153759241104126,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.03204,
+      "step": 92,
+      "tokens/total": 2784480,
+      "tokens/train_per_sec_per_gpu": 36.23,
+      "tokens/trainable": 41972
+    },
+    {
+      "epoch": 0.36328125,
+      "grad_norm": 0.5270156264305115,
+      "learning_rate": 9.586188413468492e-05,
+      "loss": 0.023751404136419296,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02404,
+      "step": 93,
+      "tokens/total": 2814928,
+      "tokens/train_per_sec_per_gpu": 35.43,
+      "tokens/trainable": 42442
+    },
+    {
+      "epoch": 0.3671875,
+      "grad_norm": 0.30221623182296753,
+      "learning_rate": 9.57394378802934e-05,
+      "loss": 0.012607569806277752,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01269,
+      "step": 94,
+      "tokens/total": 2845408,
+      "tokens/train_per_sec_per_gpu": 37.18,
+      "tokens/trainable": 42928
+    },
+    {
+      "epoch": 0.37109375,
+      "grad_norm": 0.34380587935447693,
+      "learning_rate": 9.56152962916e-05,
+      "loss": 0.021355444565415382,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02159,
+      "step": 95,
+      "tokens/total": 2875760,
+      "tokens/train_per_sec_per_gpu": 39.28,
+      "tokens/trainable": 43426
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 0.5729417204856873,
+      "learning_rate": 9.548946453464296e-05,
+      "loss": 0.011794034391641617,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01186,
+      "step": 96,
+      "tokens/total": 2906096,
+      "tokens/train_per_sec_per_gpu": 33.52,
+      "tokens/trainable": 43869
+    },
+    {
+      "epoch": 0.37890625,
+      "grad_norm": 0.1258164346218109,
+      "learning_rate": 9.53619478457953e-05,
+      "loss": 0.002208392135798931,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00221,
+      "step": 97,
+      "tokens/total": 2936560,
+      "tokens/train_per_sec_per_gpu": 32.02,
+      "tokens/trainable": 44300
+    },
+    {
+      "epoch": 0.3828125,
+      "grad_norm": 0.43118366599082947,
+      "learning_rate": 9.523275153154695e-05,
+      "loss": 0.015600357204675674,
+      "memory/device_reserved (GiB)": 35.19,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01572,
+      "step": 98,
+      "tokens/total": 2966880,
+      "tokens/train_per_sec_per_gpu": 39.37,
+      "tokens/trainable": 44791
+    },
+    {
+      "epoch": 0.38671875,
+      "grad_norm": 0.505718469619751,
+      "learning_rate": 9.51018809682839e-05,
+      "loss": 0.011874521151185036,
+      "memory/device_reserved (GiB)": 35.76,
+      "memory/max_active (GiB)": 34.03,
+      "memory/max_allocated (GiB)": 34.03,
+      "ppl": 1.01195,
+      "step": 99,
+      "tokens/total": 2997472,
+      "tokens/train_per_sec_per_gpu": 36.71,
+      "tokens/trainable": 45254
+    },
+    {
+      "epoch": 0.390625,
+      "grad_norm": 3.060197591781616,
+      "learning_rate": 9.49693416020645e-05,
+      "loss": 0.014575858600437641,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01468,
+      "step": 100,
+      "tokens/total": 3027840,
+      "tokens/train_per_sec_per_gpu": 33.57,
+      "tokens/trainable": 45698
+    },
+    {
+      "epoch": 0.39453125,
+      "grad_norm": 0.38254642486572266,
+      "learning_rate": 9.483513894839276e-05,
+      "loss": 0.009431221522390842,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00948,
+      "step": 101,
+      "tokens/total": 3058080,
+      "tokens/train_per_sec_per_gpu": 33.57,
+      "tokens/trainable": 46141
+    },
+    {
+      "epoch": 0.3984375,
+      "grad_norm": 0.8782851099967957,
+      "learning_rate": 9.469927859198888e-05,
+      "loss": 0.05128197371959686,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.05262,
+      "step": 102,
+      "tokens/total": 3088656,
+      "tokens/train_per_sec_per_gpu": 33.05,
+      "tokens/trainable": 46598
+    },
+    {
+      "epoch": 0.40234375,
+      "grad_norm": 0.42221176624298096,
+      "learning_rate": 9.456176618655689e-05,
+      "loss": 0.009157870896160603,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.0092,
+      "step": 103,
+      "tokens/total": 3119232,
+      "tokens/train_per_sec_per_gpu": 32.45,
+      "tokens/trainable": 47065
+    },
+    {
+      "epoch": 0.40625,
+      "grad_norm": 1.2625210285186768,
+      "learning_rate": 9.442260745454927e-05,
+      "loss": 0.01384878158569336,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01395,
+      "step": 104,
+      "tokens/total": 3149504,
+      "tokens/train_per_sec_per_gpu": 38.29,
+      "tokens/trainable": 47554
+    },
+    {
+      "epoch": 0.41015625,
+      "grad_norm": 0.3072832226753235,
+      "learning_rate": 9.428180818692884e-05,
+      "loss": 0.00823851116001606,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00827,
+      "step": 105,
+      "tokens/total": 3179872,
+      "tokens/train_per_sec_per_gpu": 36.88,
+      "tokens/trainable": 48047
+    },
+    {
+      "epoch": 0.4140625,
+      "grad_norm": 0.7415916323661804,
+      "learning_rate": 9.413937424292791e-05,
+      "loss": 0.03210819885134697,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.03263,
+      "step": 106,
+      "tokens/total": 3208128,
+      "tokens/train_per_sec_per_gpu": 34.62,
+      "tokens/trainable": 48498
+    },
+    {
+      "epoch": 0.41796875,
+      "grad_norm": 0.25024500489234924,
+      "learning_rate": 9.399531154980424e-05,
+      "loss": 0.00918180774897337,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.00922,
+      "step": 107,
+      "tokens/total": 3238224,
+      "tokens/train_per_sec_per_gpu": 31.18,
+      "tokens/trainable": 48933
+    },
+    {
+      "epoch": 0.421875,
+      "grad_norm": 0.5457918047904968,
+      "learning_rate": 9.384962610259455e-05,
+      "loss": 0.015052877366542816,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01517,
+      "step": 108,
+      "tokens/total": 3268592,
+      "tokens/train_per_sec_per_gpu": 38.48,
+      "tokens/trainable": 49404
+    },
+    {
+      "epoch": 0.42578125,
+      "grad_norm": 0.2508637607097626,
+      "learning_rate": 9.370232396386494e-05,
+      "loss": 0.007351120002567768,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00738,
+      "step": 109,
+      "tokens/total": 3298720,
+      "tokens/train_per_sec_per_gpu": 37.32,
+      "tokens/trainable": 49871
+    },
+    {
+      "epoch": 0.4296875,
+      "grad_norm": 0.2964157462120056,
+      "learning_rate": 9.355341126345868e-05,
+      "loss": 0.014206478372216225,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.01431,
+      "step": 110,
+      "tokens/total": 3329040,
+      "tokens/train_per_sec_per_gpu": 33.58,
+      "tokens/trainable": 50311
+    },
+    {
+      "epoch": 0.43359375,
+      "grad_norm": 0.3096546232700348,
+      "learning_rate": 9.340289419824107e-05,
+      "loss": 0.010700306855142117,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.01076,
+      "step": 111,
+      "tokens/total": 3359248,
+      "tokens/train_per_sec_per_gpu": 33.86,
+      "tokens/trainable": 50754
+    },
+    {
+      "epoch": 0.4375,
+      "grad_norm": 0.4973372220993042,
+      "learning_rate": 9.325077903184159e-05,
+      "loss": 0.013185497373342514,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01327,
+      "step": 112,
+      "tokens/total": 3389280,
+      "tokens/train_per_sec_per_gpu": 32.1,
+      "tokens/trainable": 51200
+    },
+    {
+      "epoch": 0.44140625,
+      "grad_norm": 0.4105257987976074,
+      "learning_rate": 9.30970720943932e-05,
+      "loss": 0.0069708251394331455,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.007,
+      "step": 113,
+      "tokens/total": 3419728,
+      "tokens/train_per_sec_per_gpu": 33.22,
+      "tokens/trainable": 51656
+    },
+    {
+      "epoch": 0.4453125,
+      "grad_norm": 0.2707289159297943,
+      "learning_rate": 9.2941779782269e-05,
+      "loss": 0.0033873419743031263,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.00339,
+      "step": 114,
+      "tokens/total": 3449984,
+      "tokens/train_per_sec_per_gpu": 30.92,
+      "tokens/trainable": 52069
+    },
+    {
+      "epoch": 0.44921875,
+      "grad_norm": 0.5602349042892456,
+      "learning_rate": 9.278490855781596e-05,
+      "loss": 0.00848664902150631,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00852,
+      "step": 115,
+      "tokens/total": 3480352,
+      "tokens/train_per_sec_per_gpu": 36.94,
+      "tokens/trainable": 52559
+    },
+    {
+      "epoch": 0.453125,
+      "grad_norm": 0.5224027037620544,
+      "learning_rate": 9.262646494908604e-05,
+      "loss": 0.02242768369615078,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02268,
+      "step": 116,
+      "tokens/total": 3510656,
+      "tokens/train_per_sec_per_gpu": 39.22,
+      "tokens/trainable": 53061
+    },
+    {
+      "epoch": 0.45703125,
+      "grad_norm": 1.2213423252105713,
+      "learning_rate": 9.246645554956457e-05,
+      "loss": 0.020464815199375153,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.02068,
+      "step": 117,
+      "tokens/total": 3541104,
+      "tokens/train_per_sec_per_gpu": 35.89,
+      "tokens/trainable": 53520
+    },
+    {
+      "epoch": 0.4609375,
+      "grad_norm": 1.30115807056427,
+      "learning_rate": 9.230488701789578e-05,
+      "loss": 0.025597743690013885,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.02593,
+      "step": 118,
+      "tokens/total": 3571744,
+      "tokens/train_per_sec_per_gpu": 33.36,
+      "tokens/trainable": 53976
+    },
+    {
+      "epoch": 0.46484375,
+      "grad_norm": 1.0988223552703857,
+      "learning_rate": 9.214176607760577e-05,
+      "loss": 0.02888210117816925,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.0293,
+      "step": 119,
+      "tokens/total": 3601808,
+      "tokens/train_per_sec_per_gpu": 34.93,
+      "tokens/trainable": 54408
+    },
+    {
+      "epoch": 0.46875,
+      "grad_norm": 0.3977462649345398,
+      "learning_rate": 9.197709951682268e-05,
+      "loss": 0.01678757183253765,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01693,
+      "step": 120,
+      "tokens/total": 3632096,
+      "tokens/train_per_sec_per_gpu": 37.14,
+      "tokens/trainable": 54884
+    },
+    {
+      "epoch": 0.47265625,
+      "grad_norm": 0.4872536063194275,
+      "learning_rate": 9.181089418799428e-05,
+      "loss": 0.022638533264398575,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.0229,
+      "step": 121,
+      "tokens/total": 3662416,
+      "tokens/train_per_sec_per_gpu": 33.15,
+      "tokens/trainable": 55303
+    },
+    {
+      "epoch": 0.4765625,
+      "grad_norm": 0.4663882255554199,
+      "learning_rate": 9.164315700760271e-05,
+      "loss": 0.023266036063432693,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.02354,
+      "step": 122,
+      "tokens/total": 3692992,
+      "tokens/train_per_sec_per_gpu": 31.94,
+      "tokens/trainable": 55743
+    },
+    {
+      "epoch": 0.48046875,
+      "grad_norm": 0.230746328830719,
+      "learning_rate": 9.147389495587671e-05,
+      "loss": 0.007810275070369244,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.00784,
+      "step": 123,
+      "tokens/total": 3722992,
+      "tokens/train_per_sec_per_gpu": 33.95,
+      "tokens/trainable": 56187
+    },
+    {
+      "epoch": 0.484375,
+      "grad_norm": 0.17318949103355408,
+      "learning_rate": 9.130311507650116e-05,
+      "loss": 0.006973203271627426,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.007,
+      "step": 124,
+      "tokens/total": 3753184,
+      "tokens/train_per_sec_per_gpu": 33.63,
+      "tokens/trainable": 56632
+    },
+    {
+      "epoch": 0.48828125,
+      "grad_norm": 0.19994892179965973,
+      "learning_rate": 9.113082447632394e-05,
+      "loss": 0.0068689389154314995,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00689,
+      "step": 125,
+      "tokens/total": 3783440,
+      "tokens/train_per_sec_per_gpu": 37.26,
+      "tokens/trainable": 57124
+    },
+    {
+      "epoch": 0.4921875,
+      "grad_norm": 0.11329996585845947,
+      "learning_rate": 9.09570303250602e-05,
+      "loss": 0.004266452044248581,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00428,
+      "step": 126,
+      "tokens/total": 3813840,
+      "tokens/train_per_sec_per_gpu": 33.38,
+      "tokens/trainable": 57570
+    },
+    {
+      "epoch": 0.49609375,
+      "grad_norm": 0.4156385362148285,
+      "learning_rate": 9.078173985499394e-05,
+      "loss": 0.02041183039546013,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02062,
+      "step": 127,
+      "tokens/total": 3844464,
+      "tokens/train_per_sec_per_gpu": 35.36,
+      "tokens/trainable": 58048
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.09819997847080231,
+      "learning_rate": 9.060496036067713e-05,
+      "loss": 0.0031540761701762676,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.00316,
+      "step": 128,
+      "tokens/total": 3872816,
+      "tokens/train_per_sec_per_gpu": 35.79,
+      "tokens/trainable": 58497
+    },
+    {
+      "epoch": 0.50390625,
+      "grad_norm": 0.34266355633735657,
+      "learning_rate": 9.042669919862615e-05,
+      "loss": 0.017772674560546875,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01793,
+      "step": 129,
+      "tokens/total": 3902992,
+      "tokens/train_per_sec_per_gpu": 33.44,
+      "tokens/trainable": 58947
+    },
+    {
+      "epoch": 0.5078125,
+      "grad_norm": 0.3253064751625061,
+      "learning_rate": 9.024696378701557e-05,
+      "loss": 0.018172409385442734,
+      "memory/device_reserved (GiB)": 34.74,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01834,
+      "step": 130,
+      "tokens/total": 3933456,
+      "tokens/train_per_sec_per_gpu": 37.56,
+      "tokens/trainable": 59415
+    },
+    {
+      "epoch": 0.51171875,
+      "grad_norm": 0.3167641758918762,
+      "learning_rate": 9.006576160536948e-05,
+      "loss": 0.008536357432603836,
+      "memory/device_reserved (GiB)": 34.91,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.00857,
+      "step": 131,
+      "tokens/total": 3963536,
+      "tokens/train_per_sec_per_gpu": 32.75,
+      "tokens/trainable": 59851
+    },
+    {
+      "epoch": 0.515625,
+      "grad_norm": 0.12403249740600586,
+      "learning_rate": 8.988310019425035e-05,
+      "loss": 0.0019342785235494375,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00194,
+      "step": 132,
+      "tokens/total": 3994000,
+      "tokens/train_per_sec_per_gpu": 36.25,
+      "tokens/trainable": 60325
+    },
+    {
+      "epoch": 0.51953125,
+      "grad_norm": 0.25862130522727966,
+      "learning_rate": 8.969898715494506e-05,
+      "loss": 0.01549853477627039,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01562,
+      "step": 133,
+      "tokens/total": 4024352,
+      "tokens/train_per_sec_per_gpu": 35.97,
+      "tokens/trainable": 60818
+    },
+    {
+      "epoch": 0.5234375,
+      "grad_norm": 0.3527744710445404,
+      "learning_rate": 8.951343014914869e-05,
+      "loss": 0.006827985402196646,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.00685,
+      "step": 134,
+      "tokens/total": 4052832,
+      "tokens/train_per_sec_per_gpu": 32.28,
+      "tokens/trainable": 61263
+    },
+    {
+      "epoch": 0.52734375,
+      "grad_norm": 0.3797135353088379,
+      "learning_rate": 8.932643689864568e-05,
+      "loss": 0.010647830553352833,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.0107,
+      "step": 135,
+      "tokens/total": 4083216,
+      "tokens/train_per_sec_per_gpu": 36.58,
+      "tokens/trainable": 61734
+    },
+    {
+      "epoch": 0.53125,
+      "grad_norm": 0.3589562177658081,
+      "learning_rate": 8.913801518498845e-05,
+      "loss": 0.022732166573405266,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.7,
+      "memory/max_allocated (GiB)": 33.7,
+      "ppl": 1.02299,
+      "step": 136,
+      "tokens/total": 4113392,
+      "tokens/train_per_sec_per_gpu": 38.83,
+      "tokens/trainable": 62210
+    },
+    {
+      "epoch": 0.53515625,
+      "grad_norm": 0.34385553002357483,
+      "learning_rate": 8.894817284917364e-05,
+      "loss": 0.017764274030923843,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01792,
+      "step": 137,
+      "tokens/total": 4143696,
+      "tokens/train_per_sec_per_gpu": 36.87,
+      "tokens/trainable": 62707
+    },
+    {
+      "epoch": 0.5390625,
+      "grad_norm": 0.24294431507587433,
+      "learning_rate": 8.875691779131569e-05,
+      "loss": 0.01984233781695366,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.02004,
+      "step": 138,
+      "tokens/total": 4174000,
+      "tokens/train_per_sec_per_gpu": 37.17,
+      "tokens/trainable": 63166
+    },
+    {
+      "epoch": 0.54296875,
+      "grad_norm": 0.6662021279335022,
+      "learning_rate": 8.856425797031829e-05,
+      "loss": 0.012637479230761528,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01272,
+      "step": 139,
+      "tokens/total": 4204432,
+      "tokens/train_per_sec_per_gpu": 36.96,
+      "tokens/trainable": 63646
+    },
+    {
+      "epoch": 0.546875,
+      "grad_norm": 0.26386311650276184,
+      "learning_rate": 8.837020140354295e-05,
+      "loss": 0.014607764780521393,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.42,
+      "memory/max_allocated (GiB)": 33.42,
+      "ppl": 1.01471,
+      "step": 140,
+      "tokens/total": 4232752,
+      "tokens/train_per_sec_per_gpu": 33.6,
+      "tokens/trainable": 64063
+    },
+    {
+      "epoch": 0.55078125,
+      "grad_norm": 0.2475082129240036,
+      "learning_rate": 8.817475616647554e-05,
+      "loss": 0.012658985331654549,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01274,
+      "step": 141,
+      "tokens/total": 4262976,
+      "tokens/train_per_sec_per_gpu": 35.67,
+      "tokens/trainable": 64520
+    },
+    {
+      "epoch": 0.5546875,
+      "grad_norm": 0.21145126223564148,
+      "learning_rate": 8.797793039239017e-05,
+      "loss": 0.009737009182572365,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.00978,
+      "step": 142,
+      "tokens/total": 4293520,
+      "tokens/train_per_sec_per_gpu": 34.48,
+      "tokens/trainable": 64958
+    },
+    {
+      "epoch": 0.55859375,
+      "grad_norm": 0.2296362668275833,
+      "learning_rate": 8.777973227201069e-05,
+      "loss": 0.013068155385553837,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.01315,
+      "step": 143,
+      "tokens/total": 4323520,
+      "tokens/train_per_sec_per_gpu": 38.3,
+      "tokens/trainable": 65437
+    },
+    {
+      "epoch": 0.5625,
+      "grad_norm": 0.2896386981010437,
+      "learning_rate": 8.758017005316988e-05,
+      "loss": 0.010017371736466885,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01007,
+      "step": 144,
+      "tokens/total": 4353616,
+      "tokens/train_per_sec_per_gpu": 37.25,
+      "tokens/trainable": 65903
+    },
+    {
+      "epoch": 0.56640625,
+      "grad_norm": 0.41864296793937683,
+      "learning_rate": 8.737925204046629e-05,
+      "loss": 0.025722038000822067,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.02606,
+      "step": 145,
+      "tokens/total": 4383856,
+      "tokens/train_per_sec_per_gpu": 32.08,
+      "tokens/trainable": 66334
+    },
+    {
+      "epoch": 0.5703125,
+      "grad_norm": 0.27801141142845154,
+      "learning_rate": 8.717698659491851e-05,
+      "loss": 0.017308663576841354,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01746,
+      "step": 146,
+      "tokens/total": 4414128,
+      "tokens/train_per_sec_per_gpu": 32.01,
+      "tokens/trainable": 66785
+    },
+    {
+      "epoch": 0.57421875,
+      "grad_norm": 0.575922966003418,
+      "learning_rate": 8.697338213361735e-05,
+      "loss": 0.017252381891012192,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.0174,
+      "step": 147,
+      "tokens/total": 4444384,
+      "tokens/train_per_sec_per_gpu": 35.81,
+      "tokens/trainable": 67199
+    },
+    {
+      "epoch": 0.578125,
+      "grad_norm": 0.4430011510848999,
+      "learning_rate": 8.676844712937552e-05,
+      "loss": 0.022439993917942047,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02269,
+      "step": 148,
+      "tokens/total": 4474912,
+      "tokens/train_per_sec_per_gpu": 35.64,
+      "tokens/trainable": 67683
+    },
+    {
+      "epoch": 0.58203125,
+      "grad_norm": 0.26630517840385437,
+      "learning_rate": 8.656219011037509e-05,
+      "loss": 0.010438371449708939,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01049,
+      "step": 149,
+      "tokens/total": 4505328,
+      "tokens/train_per_sec_per_gpu": 34.4,
+      "tokens/trainable": 68127
+    },
+    {
+      "epoch": 0.5859375,
+      "grad_norm": 0.15251637995243073,
+      "learning_rate": 8.63546196598125e-05,
+      "loss": 0.004219424910843372,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00423,
+      "step": 150,
+      "tokens/total": 4535424,
+      "tokens/train_per_sec_per_gpu": 33.75,
+      "tokens/trainable": 68595
+    },
+    {
+      "epoch": 0.58984375,
+      "grad_norm": 0.21238818764686584,
+      "learning_rate": 8.614574441554145e-05,
+      "loss": 0.01187204197049141,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.01194,
+      "step": 151,
+      "tokens/total": 4565536,
+      "tokens/train_per_sec_per_gpu": 30.68,
+      "tokens/trainable": 69062
+    },
+    {
+      "epoch": 0.59375,
+      "grad_norm": 0.5052844285964966,
+      "learning_rate": 8.593557306971349e-05,
+      "loss": 0.007624611258506775,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.00765,
+      "step": 152,
+      "tokens/total": 4596048,
+      "tokens/train_per_sec_per_gpu": 35.96,
+      "tokens/trainable": 69509
+    },
+    {
+      "epoch": 0.59765625,
+      "grad_norm": 0.46278703212738037,
+      "learning_rate": 8.572411436841618e-05,
+      "loss": 0.01224217563867569,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01232,
+      "step": 153,
+      "tokens/total": 4626480,
+      "tokens/train_per_sec_per_gpu": 34.45,
+      "tokens/trainable": 69972
+    },
+    {
+      "epoch": 0.6015625,
+      "grad_norm": 0.4227867126464844,
+      "learning_rate": 8.551137711130922e-05,
+      "loss": 0.0060340641066432,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00605,
+      "step": 154,
+      "tokens/total": 4656800,
+      "tokens/train_per_sec_per_gpu": 37.3,
+      "tokens/trainable": 70439
+    },
+    {
+      "epoch": 0.60546875,
+      "grad_norm": 0.5990466475486755,
+      "learning_rate": 8.529737015125824e-05,
+      "loss": 0.018272625282406807,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.01844,
+      "step": 155,
+      "tokens/total": 4685120,
+      "tokens/train_per_sec_per_gpu": 37.43,
+      "tokens/trainable": 70881
+    },
+    {
+      "epoch": 0.609375,
+      "grad_norm": 0.6471202969551086,
+      "learning_rate": 8.508210239396639e-05,
+      "loss": 0.01053079217672348,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.01059,
+      "step": 156,
+      "tokens/total": 4715664,
+      "tokens/train_per_sec_per_gpu": 34.48,
+      "tokens/trainable": 71339
+    },
+    {
+      "epoch": 0.61328125,
+      "grad_norm": 0.00898966658860445,
+      "learning_rate": 8.486558279760375e-05,
+      "loss": 0.0001828348613344133,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00018,
+      "step": 157,
+      "tokens/total": 4745824,
+      "tokens/train_per_sec_per_gpu": 35.55,
+      "tokens/trainable": 71801
+    },
+    {
+      "epoch": 0.6171875,
+      "grad_norm": 0.2965949773788452,
+      "learning_rate": 8.464782037243449e-05,
+      "loss": 0.014287484809756279,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01439,
+      "step": 158,
+      "tokens/total": 4776224,
+      "tokens/train_per_sec_per_gpu": 32.65,
+      "tokens/trainable": 72237
+    },
+    {
+      "epoch": 0.62109375,
+      "grad_norm": 0.3070797324180603,
+      "learning_rate": 8.442882418044202e-05,
+      "loss": 0.025281894952058792,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.0256,
+      "step": 159,
+      "tokens/total": 4806304,
+      "tokens/train_per_sec_per_gpu": 31.03,
+      "tokens/trainable": 72664
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 0.29432007670402527,
+      "learning_rate": 8.420860333495179e-05,
+      "loss": 0.005029833409935236,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00504,
+      "step": 160,
+      "tokens/total": 4836688,
+      "tokens/train_per_sec_per_gpu": 32.1,
+      "tokens/trainable": 73119
+    }
+  ],
+  "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": 3.282942789264799e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

aft_wave_v2/control_matched__charter2/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/control_matched__charter2/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/control_matched__charter2/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": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "gate_proj",
+    "v_proj",
+    "up_proj"
+  ],
+  "target_parameters": [],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-192/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e11c0436419262a8d7dffabe6ab36e81648ca4d1a983e89a6d8cd144e5e75faf
+size 547777976

aft_wave_v2/control_matched__charter2/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/control_matched__charter2/training/checkpoints/checkpoint-192/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:43f0be6f7706c4a9910a7d43b27df21ebe3e39389dd5b7007be5ed414249cd51
+size 1048106435

aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-192/rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3e7d3b55fe6373e1452968ec327f795eb851f8990c862c581f02fa825e7c7f48
+size 14645

aft_wave_v2/control_matched__charter2/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/control_matched__charter2/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/control_matched__charter2/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/control_matched__charter2/training/checkpoints/checkpoint-192/tokens_state.json

@@ -0,0 +1 @@
+{"total": 5807952, "trainable": 87814}

aft_wave_v2/control_matched__charter2/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": 8.088534355163574,
+      "learning_rate": 0.0,
+      "loss": 0.1503293514251709,
+      "memory/device_reserved (GiB)": 33.9,
+      "memory/max_active (GiB)": 32.79,
+      "memory/max_allocated (GiB)": 32.79,
+      "ppl": 1.16222,
+      "step": 1,
+      "tokens/total": 30176,
+      "tokens/train_per_sec_per_gpu": 28.32,
+      "tokens/trainable": 425
+    },
+    {
+      "epoch": 0.0078125,
+      "grad_norm": 0.8953529596328735,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.15210241079330444,
+      "memory/device_reserved (GiB)": 34.68,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.16428,
+      "step": 2,
+      "tokens/total": 60272,
+      "tokens/train_per_sec_per_gpu": 33.91,
+      "tokens/trainable": 872
+    },
+    {
+      "epoch": 0.01171875,
+      "grad_norm": 1.4013757705688477,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.14530743658542633,
+      "memory/device_reserved (GiB)": 34.72,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.1564,
+      "step": 3,
+      "tokens/total": 90512,
+      "tokens/train_per_sec_per_gpu": 35.5,
+      "tokens/trainable": 1352
+    },
+    {
+      "epoch": 0.015625,
+      "grad_norm": 0.9737329483032227,
+      "learning_rate": 1.2e-05,
+      "loss": 0.13504940271377563,
+      "memory/device_reserved (GiB)": 34.73,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.14459,
+      "step": 4,
+      "tokens/total": 120944,
+      "tokens/train_per_sec_per_gpu": 31.75,
+      "tokens/trainable": 1777
+    },
+    {
+      "epoch": 0.01953125,
+      "grad_norm": 1.5541704893112183,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.13814105093479156,
+      "memory/device_reserved (GiB)": 35.12,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.14814,
+      "step": 5,
+      "tokens/total": 151440,
+      "tokens/train_per_sec_per_gpu": 37.35,
+      "tokens/trainable": 2260
+    },
+    {
+      "epoch": 0.0234375,
+      "grad_norm": 6.433568477630615,
+      "learning_rate": 2e-05,
+      "loss": 0.12622183561325073,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.13453,
+      "step": 6,
+      "tokens/total": 181984,
+      "tokens/train_per_sec_per_gpu": 33.21,
+      "tokens/trainable": 2704
+    },
+    {
+      "epoch": 0.02734375,
+      "grad_norm": 1.4011743068695068,
+      "learning_rate": 2.4e-05,
+      "loss": 0.11199458688497543,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.11851,
+      "step": 7,
+      "tokens/total": 212336,
+      "tokens/train_per_sec_per_gpu": 34.37,
+      "tokens/trainable": 3143
+    },
+    {
+      "epoch": 0.03125,
+      "grad_norm": 2.038933038711548,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.09219667315483093,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.09658,
+      "step": 8,
+      "tokens/total": 242592,
+      "tokens/train_per_sec_per_gpu": 34.78,
+      "tokens/trainable": 3586
+    },
+    {
+      "epoch": 0.03515625,
+      "grad_norm": 4.82806396484375,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.04823308065533638,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.04942,
+      "step": 9,
+      "tokens/total": 272784,
+      "tokens/train_per_sec_per_gpu": 29.59,
+      "tokens/trainable": 4007
+    },
+    {
+      "epoch": 0.0390625,
+      "grad_norm": 1.3691715002059937,
+      "learning_rate": 3.6e-05,
+      "loss": 0.04049058258533478,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.04132,
+      "step": 10,
+      "tokens/total": 303184,
+      "tokens/train_per_sec_per_gpu": 39.66,
+      "tokens/trainable": 4481
+    },
+    {
+      "epoch": 0.04296875,
+      "grad_norm": 2.285440444946289,
+      "learning_rate": 4e-05,
+      "loss": 0.06339800357818604,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.06545,
+      "step": 11,
+      "tokens/total": 333296,
+      "tokens/train_per_sec_per_gpu": 35.09,
+      "tokens/trainable": 4958
+    },
+    {
+      "epoch": 0.046875,
+      "grad_norm": 5.0530476570129395,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.07100467383861542,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.07359,
+      "step": 12,
+      "tokens/total": 363840,
+      "tokens/train_per_sec_per_gpu": 34.81,
+      "tokens/trainable": 5440
+    },
+    {
+      "epoch": 0.05078125,
+      "grad_norm": 3.5007340908050537,
+      "learning_rate": 4.8e-05,
+      "loss": 0.061270229518413544,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.06319,
+      "step": 13,
+      "tokens/total": 394112,
+      "tokens/train_per_sec_per_gpu": 35.82,
+      "tokens/trainable": 5871
+    },
+    {
+      "epoch": 0.0546875,
+      "grad_norm": 2.6390483379364014,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.043673258274793625,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.04464,
+      "step": 14,
+      "tokens/total": 424656,
+      "tokens/train_per_sec_per_gpu": 35.29,
+      "tokens/trainable": 6360
+    },
+    {
+      "epoch": 0.05859375,
+      "grad_norm": 2.6278514862060547,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 0.06025860086083412,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.06211,
+      "step": 15,
+      "tokens/total": 455152,
+      "tokens/train_per_sec_per_gpu": 36.67,
+      "tokens/trainable": 6841
+    },
+    {
+      "epoch": 0.0625,
+      "grad_norm": 5.878053665161133,
+      "learning_rate": 6e-05,
+      "loss": 0.09122475981712341,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.09552,
+      "step": 16,
+      "tokens/total": 485328,
+      "tokens/train_per_sec_per_gpu": 31.7,
+      "tokens/trainable": 7254
+    },
+    {
+      "epoch": 0.06640625,
+      "grad_norm": 1.9717961549758911,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.017721600830554962,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01788,
+      "step": 17,
+      "tokens/total": 515744,
+      "tokens/train_per_sec_per_gpu": 36.52,
+      "tokens/trainable": 7723
+    },
+    {
+      "epoch": 0.0703125,
+      "grad_norm": 1.783061146736145,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.04740295931696892,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.04854,
+      "step": 18,
+      "tokens/total": 546144,
+      "tokens/train_per_sec_per_gpu": 34.38,
+      "tokens/trainable": 8208
+    },
+    {
+      "epoch": 0.07421875,
+      "grad_norm": 1.916291356086731,
+      "learning_rate": 7.2e-05,
+      "loss": 0.025028439238667488,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02534,
+      "step": 19,
+      "tokens/total": 576560,
+      "tokens/train_per_sec_per_gpu": 35.49,
+      "tokens/trainable": 8699
+    },
+    {
+      "epoch": 0.078125,
+      "grad_norm": 2.9791574478149414,
+      "learning_rate": 7.6e-05,
+      "loss": 0.055660396814346313,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.05724,
+      "step": 20,
+      "tokens/total": 607056,
+      "tokens/train_per_sec_per_gpu": 40.56,
+      "tokens/trainable": 9214
+    },
+    {
+      "epoch": 0.08203125,
+      "grad_norm": 1.2719122171401978,
+      "learning_rate": 8e-05,
+      "loss": 0.024366460740566254,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02467,
+      "step": 21,
+      "tokens/total": 637488,
+      "tokens/train_per_sec_per_gpu": 32.12,
+      "tokens/trainable": 9639
+    },
+    {
+      "epoch": 0.0859375,
+      "grad_norm": 2.241725444793701,
+      "learning_rate": 8.4e-05,
+      "loss": 0.05170319974422455,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.05306,
+      "step": 22,
+      "tokens/total": 667632,
+      "tokens/train_per_sec_per_gpu": 36.71,
+      "tokens/trainable": 10111
+    },
+    {
+      "epoch": 0.08984375,
+      "grad_norm": 2.2158987522125244,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.04347127676010132,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.04443,
+      "step": 23,
+      "tokens/total": 697872,
+      "tokens/train_per_sec_per_gpu": 33.4,
+      "tokens/trainable": 10560
+    },
+    {
+      "epoch": 0.09375,
+      "grad_norm": 1.028172254562378,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 0.016212783753871918,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01634,
+      "step": 24,
+      "tokens/total": 728432,
+      "tokens/train_per_sec_per_gpu": 32.0,
+      "tokens/trainable": 11018
+    },
+    {
+      "epoch": 0.09765625,
+      "grad_norm": 1.0408563613891602,
+      "learning_rate": 9.6e-05,
+      "loss": 0.023505806922912598,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02378,
+      "step": 25,
+      "tokens/total": 756656,
+      "tokens/train_per_sec_per_gpu": 40.53,
+      "tokens/trainable": 11442
+    },
+    {
+      "epoch": 0.1015625,
+      "grad_norm": 1.6656584739685059,
+      "learning_rate": 0.0001,
+      "loss": 0.05568547546863556,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.05727,
+      "step": 26,
+      "tokens/total": 787120,
+      "tokens/train_per_sec_per_gpu": 34.83,
+      "tokens/trainable": 11899
+    },
+    {
+      "epoch": 0.10546875,
+      "grad_norm": 0.7106256484985352,
+      "learning_rate": 9.99990636831587e-05,
+      "loss": 0.009144511073827744,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.00919,
+      "step": 27,
+      "tokens/total": 817504,
+      "tokens/train_per_sec_per_gpu": 32.1,
+      "tokens/trainable": 12341
+    },
+    {
+      "epoch": 0.109375,
+      "grad_norm": 0.8463047742843628,
+      "learning_rate": 9.999625477159879e-05,
+      "loss": 0.020250165835022926,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02046,
+      "step": 28,
+      "tokens/total": 848128,
+      "tokens/train_per_sec_per_gpu": 38.39,
+      "tokens/trainable": 12825
+    },
+    {
+      "epoch": 0.11328125,
+      "grad_norm": 1.0744967460632324,
+      "learning_rate": 9.999157338221051e-05,
+      "loss": 0.023447610437870026,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02372,
+      "step": 29,
+      "tokens/total": 878448,
+      "tokens/train_per_sec_per_gpu": 36.01,
+      "tokens/trainable": 13295
+    },
+    {
+      "epoch": 0.1171875,
+      "grad_norm": 1.4234495162963867,
+      "learning_rate": 9.998501970980562e-05,
+      "loss": 0.03458714857697487,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.03519,
+      "step": 30,
+      "tokens/total": 909008,
+      "tokens/train_per_sec_per_gpu": 35.8,
+      "tokens/trainable": 13784
+    },
+    {
+      "epoch": 0.12109375,
+      "grad_norm": 0.9880445003509521,
+      "learning_rate": 9.997659402710915e-05,
+      "loss": 0.02479860559105873,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02511,
+      "step": 31,
+      "tokens/total": 939312,
+      "tokens/train_per_sec_per_gpu": 34.58,
+      "tokens/trainable": 14250
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 1.1164203882217407,
+      "learning_rate": 9.996629668474818e-05,
+      "loss": 0.023022079840302467,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02329,
+      "step": 32,
+      "tokens/total": 969264,
+      "tokens/train_per_sec_per_gpu": 35.0,
+      "tokens/trainable": 14680
+    },
+    {
+      "epoch": 0.12890625,
+      "grad_norm": 1.324739694595337,
+      "learning_rate": 9.995412811123711e-05,
+      "loss": 0.0630958303809166,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.06513,
+      "step": 33,
+      "tokens/total": 999792,
+      "tokens/train_per_sec_per_gpu": 33.83,
+      "tokens/trainable": 15112
+    },
+    {
+      "epoch": 0.1328125,
+      "grad_norm": 1.0034446716308594,
+      "learning_rate": 9.994008881295999e-05,
+      "loss": 0.04252947121858597,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.04345,
+      "step": 34,
+      "tokens/total": 1029840,
+      "tokens/train_per_sec_per_gpu": 34.46,
+      "tokens/trainable": 15573
+    },
+    {
+      "epoch": 0.13671875,
+      "grad_norm": 1.1391361951828003,
+      "learning_rate": 9.992417937414932e-05,
+      "loss": 0.0361175499856472,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03678,
+      "step": 35,
+      "tokens/total": 1060416,
+      "tokens/train_per_sec_per_gpu": 34.24,
+      "tokens/trainable": 16062
+    },
+    {
+      "epoch": 0.140625,
+      "grad_norm": 0.48655080795288086,
+      "learning_rate": 9.99064004568618e-05,
+      "loss": 0.015174772590398788,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.72,
+      "memory/max_allocated (GiB)": 33.72,
+      "ppl": 1.01529,
+      "step": 36,
+      "tokens/total": 1090464,
+      "tokens/train_per_sec_per_gpu": 31.53,
+      "tokens/trainable": 16500
+    },
+    {
+      "epoch": 0.14453125,
+      "grad_norm": 0.7631387114524841,
+      "learning_rate": 9.988675280095074e-05,
+      "loss": 0.04621092230081558,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.0473,
+      "step": 37,
+      "tokens/total": 1120944,
+      "tokens/train_per_sec_per_gpu": 34.9,
+      "tokens/trainable": 16936
+    },
+    {
+      "epoch": 0.1484375,
+      "grad_norm": 1.1162818670272827,
+      "learning_rate": 9.986523722403528e-05,
+      "loss": 0.046002037823200226,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.04708,
+      "step": 38,
+      "tokens/total": 1151360,
+      "tokens/train_per_sec_per_gpu": 29.32,
+      "tokens/trainable": 17357
+    },
+    {
+      "epoch": 0.15234375,
+      "grad_norm": 4.360172271728516,
+      "learning_rate": 9.984185462146642e-05,
+      "loss": 0.052004992961883545,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.05338,
+      "step": 39,
+      "tokens/total": 1181728,
+      "tokens/train_per_sec_per_gpu": 36.82,
+      "tokens/trainable": 17831
+    },
+    {
+      "epoch": 0.15625,
+      "grad_norm": 0.8433209657669067,
+      "learning_rate": 9.98166059662897e-05,
+      "loss": 0.04868919029831886,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.04989,
+      "step": 40,
+      "tokens/total": 1212064,
+      "tokens/train_per_sec_per_gpu": 36.15,
+      "tokens/trainable": 18315
+    },
+    {
+      "epoch": 0.16015625,
+      "grad_norm": 0.9528040289878845,
+      "learning_rate": 9.978949230920472e-05,
+      "loss": 0.029145417734980583,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.02957,
+      "step": 41,
+      "tokens/total": 1242592,
+      "tokens/train_per_sec_per_gpu": 32.49,
+      "tokens/trainable": 18772
+    },
+    {
+      "epoch": 0.1640625,
+      "grad_norm": 0.5701055526733398,
+      "learning_rate": 9.976051477852141e-05,
+      "loss": 0.01586967147886753,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.016,
+      "step": 42,
+      "tokens/total": 1272800,
+      "tokens/train_per_sec_per_gpu": 32.32,
+      "tokens/trainable": 19197
+    },
+    {
+      "epoch": 0.16796875,
+      "grad_norm": 0.372957706451416,
+      "learning_rate": 9.972967458011312e-05,
+      "loss": 0.01052568107843399,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 34.04,
+      "memory/max_allocated (GiB)": 34.04,
+      "ppl": 1.01058,
+      "step": 43,
+      "tokens/total": 1303600,
+      "tokens/train_per_sec_per_gpu": 31.68,
+      "tokens/trainable": 19647
+    },
+    {
+      "epoch": 0.171875,
+      "grad_norm": 0.550298810005188,
+      "learning_rate": 9.96969729973664e-05,
+      "loss": 0.019961029291152954,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02016,
+      "step": 44,
+      "tokens/total": 1333936,
+      "tokens/train_per_sec_per_gpu": 33.83,
+      "tokens/trainable": 20097
+    },
+    {
+      "epoch": 0.17578125,
+      "grad_norm": 3.7621572017669678,
+      "learning_rate": 9.966241139112754e-05,
+      "loss": 0.06465393304824829,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.06679,
+      "step": 45,
+      "tokens/total": 1364288,
+      "tokens/train_per_sec_per_gpu": 31.92,
+      "tokens/trainable": 20529
+    },
+    {
+      "epoch": 0.1796875,
+      "grad_norm": 1.1490390300750732,
+      "learning_rate": 9.96259911996461e-05,
+      "loss": 0.055192310363054276,
+      "memory/device_reserved (GiB)": 36.04,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.05674,
+      "step": 46,
+      "tokens/total": 1394432,
+      "tokens/train_per_sec_per_gpu": 34.68,
+      "tokens/trainable": 20973
+    },
+    {
+      "epoch": 0.18359375,
+      "grad_norm": 0.5583573579788208,
+      "learning_rate": 9.958771393851491e-05,
+      "loss": 0.024018850177526474,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.98,
+      "memory/max_allocated (GiB)": 33.98,
+      "ppl": 1.02431,
+      "step": 47,
+      "tokens/total": 1425088,
+      "tokens/train_per_sec_per_gpu": 35.36,
+      "tokens/trainable": 21437
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 0.7726427912712097,
+      "learning_rate": 9.954758120060702e-05,
+      "loss": 0.021447816863656044,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.02168,
+      "step": 48,
+      "tokens/total": 1455456,
+      "tokens/train_per_sec_per_gpu": 35.57,
+      "tokens/trainable": 21860
+    },
+    {
+      "epoch": 0.19140625,
+      "grad_norm": 0.9507198929786682,
+      "learning_rate": 9.950559465600948e-05,
+      "loss": 0.03307538479566574,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.7,
+      "memory/max_allocated (GiB)": 33.7,
+      "ppl": 1.03363,
+      "step": 49,
+      "tokens/total": 1485376,
+      "tokens/train_per_sec_per_gpu": 30.1,
+      "tokens/trainable": 22273
+    },
+    {
+      "epoch": 0.1953125,
+      "grad_norm": 0.5201236009597778,
+      "learning_rate": 9.946175605195379e-05,
+      "loss": 0.019475962966680527,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01967,
+      "step": 50,
+      "tokens/total": 1515744,
+      "tokens/train_per_sec_per_gpu": 36.83,
+      "tokens/trainable": 22720
+    },
+    {
+      "epoch": 0.19921875,
+      "grad_norm": 0.5436747670173645,
+      "learning_rate": 9.941606721274322e-05,
+      "loss": 0.021235385909676552,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.02146,
+      "step": 51,
+      "tokens/total": 1545984,
+      "tokens/train_per_sec_per_gpu": 32.55,
+      "tokens/trainable": 23168
+    },
+    {
+      "epoch": 0.203125,
+      "grad_norm": 0.47054970264434814,
+      "learning_rate": 9.936853003967685e-05,
+      "loss": 0.02223706617951393,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.02249,
+      "step": 52,
+      "tokens/total": 1576288,
+      "tokens/train_per_sec_per_gpu": 37.44,
+      "tokens/trainable": 23663
+    },
+    {
+      "epoch": 0.20703125,
+      "grad_norm": 12.063824653625488,
+      "learning_rate": 9.93191465109705e-05,
+      "loss": 0.01991445943713188,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02011,
+      "step": 53,
+      "tokens/total": 1606528,
+      "tokens/train_per_sec_per_gpu": 34.49,
+      "tokens/trainable": 24128
+    },
+    {
+      "epoch": 0.2109375,
+      "grad_norm": 0.731513500213623,
+      "learning_rate": 9.926791868167438e-05,
+      "loss": 0.035772789269685745,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.66,
+      "memory/max_allocated (GiB)": 33.66,
+      "ppl": 1.03642,
+      "step": 54,
+      "tokens/total": 1636400,
+      "tokens/train_per_sec_per_gpu": 38.33,
+      "tokens/trainable": 24601
+    },
+    {
+      "epoch": 0.21484375,
+      "grad_norm": 0.8308107256889343,
+      "learning_rate": 9.921484868358753e-05,
+      "loss": 0.04387127608060837,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.04485,
+      "step": 55,
+      "tokens/total": 1666896,
+      "tokens/train_per_sec_per_gpu": 34.44,
+      "tokens/trainable": 25099
+    },
+    {
+      "epoch": 0.21875,
+      "grad_norm": 0.3609774708747864,
+      "learning_rate": 9.915993872516924e-05,
+      "loss": 0.009335450828075409,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00938,
+      "step": 56,
+      "tokens/total": 1697232,
+      "tokens/train_per_sec_per_gpu": 37.08,
+      "tokens/trainable": 25573
+    },
+    {
+      "epoch": 0.22265625,
+      "grad_norm": 0.5462154150009155,
+      "learning_rate": 9.9103191091447e-05,
+      "loss": 0.016265662387013435,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.0164,
+      "step": 57,
+      "tokens/total": 1727440,
+      "tokens/train_per_sec_per_gpu": 33.69,
+      "tokens/trainable": 25997
+    },
+    {
+      "epoch": 0.2265625,
+      "grad_norm": 3.2781224250793457,
+      "learning_rate": 9.904460814392147e-05,
+      "loss": 0.015918206423521042,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.01605,
+      "step": 58,
+      "tokens/total": 1757952,
+      "tokens/train_per_sec_per_gpu": 34.37,
+      "tokens/trainable": 26441
+    },
+    {
+      "epoch": 0.23046875,
+      "grad_norm": 1.6426713466644287,
+      "learning_rate": 9.898419232046825e-05,
+      "loss": 0.0164799727499485,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01662,
+      "step": 59,
+      "tokens/total": 1788576,
+      "tokens/train_per_sec_per_gpu": 34.54,
+      "tokens/trainable": 26915
+    },
+    {
+      "epoch": 0.234375,
+      "grad_norm": 0.5991567969322205,
+      "learning_rate": 9.892194613523633e-05,
+      "loss": 0.010309271514415741,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01036,
+      "step": 60,
+      "tokens/total": 1818816,
+      "tokens/train_per_sec_per_gpu": 34.2,
+      "tokens/trainable": 27355
+    },
+    {
+      "epoch": 0.23828125,
+      "grad_norm": 0.12077115476131439,
+      "learning_rate": 9.885787217854357e-05,
+      "loss": 0.00214880402199924,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00215,
+      "step": 61,
+      "tokens/total": 1849424,
+      "tokens/train_per_sec_per_gpu": 39.71,
+      "tokens/trainable": 27862
+    },
+    {
+      "epoch": 0.2421875,
+      "grad_norm": 0.8362685441970825,
+      "learning_rate": 9.879197311676887e-05,
+      "loss": 0.04497821629047394,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 34.02,
+      "memory/max_allocated (GiB)": 34.02,
+      "ppl": 1.04601,
+      "step": 62,
+      "tokens/total": 1880032,
+      "tokens/train_per_sec_per_gpu": 32.8,
+      "tokens/trainable": 28313
+    },
+    {
+      "epoch": 0.24609375,
+      "grad_norm": 1.2074859142303467,
+      "learning_rate": 9.872425169224113e-05,
+      "loss": 0.04683143272995949,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.04795,
+      "step": 63,
+      "tokens/total": 1910512,
+      "tokens/train_per_sec_per_gpu": 32.7,
+      "tokens/trainable": 28774
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8954252600669861,
+      "learning_rate": 9.865471072312528e-05,
+      "loss": 0.028329109773039818,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02873,
+      "step": 64,
+      "tokens/total": 1940608,
+      "tokens/train_per_sec_per_gpu": 34.07,
+      "tokens/trainable": 29222
+    },
+    {
+      "epoch": 0.25390625,
+      "grad_norm": 0.710774838924408,
+      "learning_rate": 9.858335310330492e-05,
+      "loss": 0.018383217975497246,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.01855,
+      "step": 65,
+      "tokens/total": 1971104,
+      "tokens/train_per_sec_per_gpu": 32.0,
+      "tokens/trainable": 29687
+    },
+    {
+      "epoch": 0.2578125,
+      "grad_norm": 0.841823935508728,
+      "learning_rate": 9.851018180226185e-05,
+      "loss": 0.024537263438105583,
+      "memory/device_reserved (GiB)": 35.06,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02484,
+      "step": 66,
+      "tokens/total": 2001472,
+      "tokens/train_per_sec_per_gpu": 33.93,
+      "tokens/trainable": 30114
+    },
+    {
+      "epoch": 0.26171875,
+      "grad_norm": 0.5852411389350891,
+      "learning_rate": 9.843519986495259e-05,
+      "loss": 0.03270117565989494,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.03324,
+      "step": 67,
+      "tokens/total": 2029696,
+      "tokens/train_per_sec_per_gpu": 38.04,
+      "tokens/trainable": 30594
+    },
+    {
+      "epoch": 0.265625,
+      "grad_norm": 0.3637949824333191,
+      "learning_rate": 9.835841041168162e-05,
+      "loss": 0.011452089995145798,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01152,
+      "step": 68,
+      "tokens/total": 2060144,
+      "tokens/train_per_sec_per_gpu": 35.62,
+      "tokens/trainable": 31073
+    },
+    {
+      "epoch": 0.26953125,
+      "grad_norm": 0.4834350049495697,
+      "learning_rate": 9.82798166379715e-05,
+      "loss": 0.032057274132966995,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.03258,
+      "step": 69,
+      "tokens/total": 2090448,
+      "tokens/train_per_sec_per_gpu": 34.38,
+      "tokens/trainable": 31553
+    },
+    {
+      "epoch": 0.2734375,
+      "grad_norm": 0.5799927115440369,
+      "learning_rate": 9.819942181443002e-05,
+      "loss": 0.016432201489806175,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01657,
+      "step": 70,
+      "tokens/total": 2120608,
+      "tokens/train_per_sec_per_gpu": 32.72,
+      "tokens/trainable": 32004
+    },
+    {
+      "epoch": 0.27734375,
+      "grad_norm": 0.6350996494293213,
+      "learning_rate": 9.811722928661392e-05,
+      "loss": 0.04276867210865021,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.35,
+      "memory/max_allocated (GiB)": 33.35,
+      "ppl": 1.0437,
+      "step": 71,
+      "tokens/total": 2148848,
+      "tokens/train_per_sec_per_gpu": 32.26,
+      "tokens/trainable": 32419
+    },
+    {
+      "epoch": 0.28125,
+      "grad_norm": 0.2791362702846527,
+      "learning_rate": 9.803324247488975e-05,
+      "loss": 0.009182551875710487,
+      "memory/device_reserved (GiB)": 35.65,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00922,
+      "step": 72,
+      "tokens/total": 2179200,
+      "tokens/train_per_sec_per_gpu": 34.98,
+      "tokens/trainable": 32878
+    },
+    {
+      "epoch": 0.28515625,
+      "grad_norm": 0.7223671674728394,
+      "learning_rate": 9.794746487429161e-05,
+      "loss": 0.028303124010562897,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.02871,
+      "step": 73,
+      "tokens/total": 2209712,
+      "tokens/train_per_sec_per_gpu": 36.21,
+      "tokens/trainable": 33348
+    },
+    {
+      "epoch": 0.2890625,
+      "grad_norm": 0.6146635413169861,
+      "learning_rate": 9.785990005437554e-05,
+      "loss": 0.010147360153496265,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.0102,
+      "step": 74,
+      "tokens/total": 2240240,
+      "tokens/train_per_sec_per_gpu": 32.86,
+      "tokens/trainable": 33815
+    },
+    {
+      "epoch": 0.29296875,
+      "grad_norm": 0.28880107402801514,
+      "learning_rate": 9.777055165907117e-05,
+      "loss": 0.012021517381072044,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01209,
+      "step": 75,
+      "tokens/total": 2270336,
+      "tokens/train_per_sec_per_gpu": 34.86,
+      "tokens/trainable": 34229
+    },
+    {
+      "epoch": 0.296875,
+      "grad_norm": 0.4516862630844116,
+      "learning_rate": 9.767942340652993e-05,
+      "loss": 0.023413028568029404,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.02369,
+      "step": 76,
+      "tokens/total": 2300784,
+      "tokens/train_per_sec_per_gpu": 36.53,
+      "tokens/trainable": 34718
+    },
+    {
+      "epoch": 0.30078125,
+      "grad_norm": 0.9154849648475647,
+      "learning_rate": 9.758651908897035e-05,
+      "loss": 0.052833881229162216,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.76,
+      "memory/max_allocated (GiB)": 33.76,
+      "ppl": 1.05425,
+      "step": 77,
+      "tokens/total": 2330880,
+      "tokens/train_per_sec_per_gpu": 32.14,
+      "tokens/trainable": 35146
+    },
+    {
+      "epoch": 0.3046875,
+      "grad_norm": 0.32223156094551086,
+      "learning_rate": 9.749184257252033e-05,
+      "loss": 0.009072870016098022,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.00911,
+      "step": 78,
+      "tokens/total": 2361392,
+      "tokens/train_per_sec_per_gpu": 35.29,
+      "tokens/trainable": 35610
+    },
+    {
+      "epoch": 0.30859375,
+      "grad_norm": 0.4267147183418274,
+      "learning_rate": 9.739539779705614e-05,
+      "loss": 0.024432381615042686,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02473,
+      "step": 79,
+      "tokens/total": 2391744,
+      "tokens/train_per_sec_per_gpu": 29.95,
+      "tokens/trainable": 36037
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 0.3152504563331604,
+      "learning_rate": 9.729718877603861e-05,
+      "loss": 0.015125786885619164,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01524,
+      "step": 80,
+      "tokens/total": 2422240,
+      "tokens/train_per_sec_per_gpu": 38.73,
+      "tokens/trainable": 36522
+    },
+    {
+      "epoch": 0.31640625,
+      "grad_norm": 0.6696128249168396,
+      "learning_rate": 9.719721959634592e-05,
+      "loss": 0.023452740162611008,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.02373,
+      "step": 81,
+      "tokens/total": 2452560,
+      "tokens/train_per_sec_per_gpu": 33.67,
+      "tokens/trainable": 36979
+    },
+    {
+      "epoch": 0.3203125,
+      "grad_norm": 0.3085023760795593,
+      "learning_rate": 9.709549441810375e-05,
+      "loss": 0.011503729969263077,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.44,
+      "memory/max_allocated (GiB)": 33.44,
+      "ppl": 1.01157,
+      "step": 82,
+      "tokens/total": 2481024,
+      "tokens/train_per_sec_per_gpu": 33.93,
+      "tokens/trainable": 37439
+    },
+    {
+      "epoch": 0.32421875,
+      "grad_norm": 0.44436803460121155,
+      "learning_rate": 9.699201747451195e-05,
+      "loss": 0.01643884740769863,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01657,
+      "step": 83,
+      "tokens/total": 2511296,
+      "tokens/train_per_sec_per_gpu": 33.54,
+      "tokens/trainable": 37892
+    },
+    {
+      "epoch": 0.328125,
+      "grad_norm": 1.1241978406906128,
+      "learning_rate": 9.688679307166854e-05,
+      "loss": 0.0325784869492054,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.03311,
+      "step": 84,
+      "tokens/total": 2541792,
+      "tokens/train_per_sec_per_gpu": 33.19,
+      "tokens/trainable": 38355
+    },
+    {
+      "epoch": 0.33203125,
+      "grad_norm": 0.493679016828537,
+      "learning_rate": 9.677982558839042e-05,
+      "loss": 0.0119265615940094,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.69,
+      "memory/max_allocated (GiB)": 33.69,
+      "ppl": 1.012,
+      "step": 85,
+      "tokens/total": 2572032,
+      "tokens/train_per_sec_per_gpu": 34.07,
+      "tokens/trainable": 38810
+    },
+    {
+      "epoch": 0.3359375,
+      "grad_norm": 1.3241630792617798,
+      "learning_rate": 9.66711194760312e-05,
+      "loss": 0.04201122000813484,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.04291,
+      "step": 86,
+      "tokens/total": 2602512,
+      "tokens/train_per_sec_per_gpu": 30.89,
+      "tokens/trainable": 39240
+    },
+    {
+      "epoch": 0.33984375,
+      "grad_norm": 0.4352894723415375,
+      "learning_rate": 9.656067925829593e-05,
+      "loss": 0.028608884662389755,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.02902,
+      "step": 87,
+      "tokens/total": 2632880,
+      "tokens/train_per_sec_per_gpu": 37.0,
+      "tokens/trainable": 39748
+    },
+    {
+      "epoch": 0.34375,
+      "grad_norm": 0.48992764949798584,
+      "learning_rate": 9.644850953105288e-05,
+      "loss": 0.035998307168483734,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.03665,
+      "step": 88,
+      "tokens/total": 2663040,
+      "tokens/train_per_sec_per_gpu": 33.35,
+      "tokens/trainable": 40188
+    },
+    {
+      "epoch": 0.34765625,
+      "grad_norm": 0.491361141204834,
+      "learning_rate": 9.633461496214225e-05,
+      "loss": 0.02381829358637333,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0241,
+      "step": 89,
+      "tokens/total": 2693504,
+      "tokens/train_per_sec_per_gpu": 33.36,
+      "tokens/trainable": 40632
+    },
+    {
+      "epoch": 0.3515625,
+      "grad_norm": 0.4422471225261688,
+      "learning_rate": 9.621900029118195e-05,
+      "loss": 0.022936182096600533,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.0232,
+      "step": 90,
+      "tokens/total": 2723792,
+      "tokens/train_per_sec_per_gpu": 31.37,
+      "tokens/trainable": 41050
+    },
+    {
+      "epoch": 0.35546875,
+      "grad_norm": 0.4762331247329712,
+      "learning_rate": 9.610167032937036e-05,
+      "loss": 0.017587462440133095,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01774,
+      "step": 91,
+      "tokens/total": 2754048,
+      "tokens/train_per_sec_per_gpu": 37.05,
+      "tokens/trainable": 41515
+    },
+    {
+      "epoch": 0.359375,
+      "grad_norm": 0.4137004017829895,
+      "learning_rate": 9.598262995928611e-05,
+      "loss": 0.03153759241104126,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.03204,
+      "step": 92,
+      "tokens/total": 2784480,
+      "tokens/train_per_sec_per_gpu": 36.23,
+      "tokens/trainable": 41972
+    },
+    {
+      "epoch": 0.36328125,
+      "grad_norm": 0.5270156264305115,
+      "learning_rate": 9.586188413468492e-05,
+      "loss": 0.023751404136419296,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.02404,
+      "step": 93,
+      "tokens/total": 2814928,
+      "tokens/train_per_sec_per_gpu": 35.43,
+      "tokens/trainable": 42442
+    },
+    {
+      "epoch": 0.3671875,
+      "grad_norm": 0.30221623182296753,
+      "learning_rate": 9.57394378802934e-05,
+      "loss": 0.012607569806277752,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01269,
+      "step": 94,
+      "tokens/total": 2845408,
+      "tokens/train_per_sec_per_gpu": 37.18,
+      "tokens/trainable": 42928
+    },
+    {
+      "epoch": 0.37109375,
+      "grad_norm": 0.34380587935447693,
+      "learning_rate": 9.56152962916e-05,
+      "loss": 0.021355444565415382,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02159,
+      "step": 95,
+      "tokens/total": 2875760,
+      "tokens/train_per_sec_per_gpu": 39.28,
+      "tokens/trainable": 43426
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 0.5729417204856873,
+      "learning_rate": 9.548946453464296e-05,
+      "loss": 0.011794034391641617,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01186,
+      "step": 96,
+      "tokens/total": 2906096,
+      "tokens/train_per_sec_per_gpu": 33.52,
+      "tokens/trainable": 43869
+    },
+    {
+      "epoch": 0.37890625,
+      "grad_norm": 0.1258164346218109,
+      "learning_rate": 9.53619478457953e-05,
+      "loss": 0.002208392135798931,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00221,
+      "step": 97,
+      "tokens/total": 2936560,
+      "tokens/train_per_sec_per_gpu": 32.02,
+      "tokens/trainable": 44300
+    },
+    {
+      "epoch": 0.3828125,
+      "grad_norm": 0.43118366599082947,
+      "learning_rate": 9.523275153154695e-05,
+      "loss": 0.015600357204675674,
+      "memory/device_reserved (GiB)": 35.19,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01572,
+      "step": 98,
+      "tokens/total": 2966880,
+      "tokens/train_per_sec_per_gpu": 39.37,
+      "tokens/trainable": 44791
+    },
+    {
+      "epoch": 0.38671875,
+      "grad_norm": 0.505718469619751,
+      "learning_rate": 9.51018809682839e-05,
+      "loss": 0.011874521151185036,
+      "memory/device_reserved (GiB)": 35.76,
+      "memory/max_active (GiB)": 34.03,
+      "memory/max_allocated (GiB)": 34.03,
+      "ppl": 1.01195,
+      "step": 99,
+      "tokens/total": 2997472,
+      "tokens/train_per_sec_per_gpu": 36.71,
+      "tokens/trainable": 45254
+    },
+    {
+      "epoch": 0.390625,
+      "grad_norm": 3.060197591781616,
+      "learning_rate": 9.49693416020645e-05,
+      "loss": 0.014575858600437641,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01468,
+      "step": 100,
+      "tokens/total": 3027840,
+      "tokens/train_per_sec_per_gpu": 33.57,
+      "tokens/trainable": 45698
+    },
+    {
+      "epoch": 0.39453125,
+      "grad_norm": 0.38254642486572266,
+      "learning_rate": 9.483513894839276e-05,
+      "loss": 0.009431221522390842,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00948,
+      "step": 101,
+      "tokens/total": 3058080,
+      "tokens/train_per_sec_per_gpu": 33.57,
+      "tokens/trainable": 46141
+    },
+    {
+      "epoch": 0.3984375,
+      "grad_norm": 0.8782851099967957,
+      "learning_rate": 9.469927859198888e-05,
+      "loss": 0.05128197371959686,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.05262,
+      "step": 102,
+      "tokens/total": 3088656,
+      "tokens/train_per_sec_per_gpu": 33.05,
+      "tokens/trainable": 46598
+    },
+    {
+      "epoch": 0.40234375,
+      "grad_norm": 0.42221176624298096,
+      "learning_rate": 9.456176618655689e-05,
+      "loss": 0.009157870896160603,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.0092,
+      "step": 103,
+      "tokens/total": 3119232,
+      "tokens/train_per_sec_per_gpu": 32.45,
+      "tokens/trainable": 47065
+    },
+    {
+      "epoch": 0.40625,
+      "grad_norm": 1.2625210285186768,
+      "learning_rate": 9.442260745454927e-05,
+      "loss": 0.01384878158569336,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01395,
+      "step": 104,
+      "tokens/total": 3149504,
+      "tokens/train_per_sec_per_gpu": 38.29,
+      "tokens/trainable": 47554
+    },
+    {
+      "epoch": 0.41015625,
+      "grad_norm": 0.3072832226753235,
+      "learning_rate": 9.428180818692884e-05,
+      "loss": 0.00823851116001606,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00827,
+      "step": 105,
+      "tokens/total": 3179872,
+      "tokens/train_per_sec_per_gpu": 36.88,
+      "tokens/trainable": 48047
+    },
+    {
+      "epoch": 0.4140625,
+      "grad_norm": 0.7415916323661804,
+      "learning_rate": 9.413937424292791e-05,
+      "loss": 0.03210819885134697,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.03263,
+      "step": 106,
+      "tokens/total": 3208128,
+      "tokens/train_per_sec_per_gpu": 34.62,
+      "tokens/trainable": 48498
+    },
+    {
+      "epoch": 0.41796875,
+      "grad_norm": 0.25024500489234924,
+      "learning_rate": 9.399531154980424e-05,
+      "loss": 0.00918180774897337,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.00922,
+      "step": 107,
+      "tokens/total": 3238224,
+      "tokens/train_per_sec_per_gpu": 31.18,
+      "tokens/trainable": 48933
+    },
+    {
+      "epoch": 0.421875,
+      "grad_norm": 0.5457918047904968,
+      "learning_rate": 9.384962610259455e-05,
+      "loss": 0.015052877366542816,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01517,
+      "step": 108,
+      "tokens/total": 3268592,
+      "tokens/train_per_sec_per_gpu": 38.48,
+      "tokens/trainable": 49404
+    },
+    {
+      "epoch": 0.42578125,
+      "grad_norm": 0.2508637607097626,
+      "learning_rate": 9.370232396386494e-05,
+      "loss": 0.007351120002567768,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00738,
+      "step": 109,
+      "tokens/total": 3298720,
+      "tokens/train_per_sec_per_gpu": 37.32,
+      "tokens/trainable": 49871
+    },
+    {
+      "epoch": 0.4296875,
+      "grad_norm": 0.2964157462120056,
+      "learning_rate": 9.355341126345868e-05,
+      "loss": 0.014206478372216225,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.01431,
+      "step": 110,
+      "tokens/total": 3329040,
+      "tokens/train_per_sec_per_gpu": 33.58,
+      "tokens/trainable": 50311
+    },
+    {
+      "epoch": 0.43359375,
+      "grad_norm": 0.3096546232700348,
+      "learning_rate": 9.340289419824107e-05,
+      "loss": 0.010700306855142117,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.01076,
+      "step": 111,
+      "tokens/total": 3359248,
+      "tokens/train_per_sec_per_gpu": 33.86,
+      "tokens/trainable": 50754
+    },
+    {
+      "epoch": 0.4375,
+      "grad_norm": 0.4973372220993042,
+      "learning_rate": 9.325077903184159e-05,
+      "loss": 0.013185497373342514,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01327,
+      "step": 112,
+      "tokens/total": 3389280,
+      "tokens/train_per_sec_per_gpu": 32.1,
+      "tokens/trainable": 51200
+    },
+    {
+      "epoch": 0.44140625,
+      "grad_norm": 0.4105257987976074,
+      "learning_rate": 9.30970720943932e-05,
+      "loss": 0.0069708251394331455,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.007,
+      "step": 113,
+      "tokens/total": 3419728,
+      "tokens/train_per_sec_per_gpu": 33.22,
+      "tokens/trainable": 51656
+    },
+    {
+      "epoch": 0.4453125,
+      "grad_norm": 0.2707289159297943,
+      "learning_rate": 9.2941779782269e-05,
+      "loss": 0.0033873419743031263,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.00339,
+      "step": 114,
+      "tokens/total": 3449984,
+      "tokens/train_per_sec_per_gpu": 30.92,
+      "tokens/trainable": 52069
+    },
+    {
+      "epoch": 0.44921875,
+      "grad_norm": 0.5602349042892456,
+      "learning_rate": 9.278490855781596e-05,
+      "loss": 0.00848664902150631,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00852,
+      "step": 115,
+      "tokens/total": 3480352,
+      "tokens/train_per_sec_per_gpu": 36.94,
+      "tokens/trainable": 52559
+    },
+    {
+      "epoch": 0.453125,
+      "grad_norm": 0.5224027037620544,
+      "learning_rate": 9.262646494908604e-05,
+      "loss": 0.02242768369615078,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02268,
+      "step": 116,
+      "tokens/total": 3510656,
+      "tokens/train_per_sec_per_gpu": 39.22,
+      "tokens/trainable": 53061
+    },
+    {
+      "epoch": 0.45703125,
+      "grad_norm": 1.2213423252105713,
+      "learning_rate": 9.246645554956457e-05,
+      "loss": 0.020464815199375153,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.02068,
+      "step": 117,
+      "tokens/total": 3541104,
+      "tokens/train_per_sec_per_gpu": 35.89,
+      "tokens/trainable": 53520
+    },
+    {
+      "epoch": 0.4609375,
+      "grad_norm": 1.30115807056427,
+      "learning_rate": 9.230488701789578e-05,
+      "loss": 0.025597743690013885,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.02593,
+      "step": 118,
+      "tokens/total": 3571744,
+      "tokens/train_per_sec_per_gpu": 33.36,
+      "tokens/trainable": 53976
+    },
+    {
+      "epoch": 0.46484375,
+      "grad_norm": 1.0988223552703857,
+      "learning_rate": 9.214176607760577e-05,
+      "loss": 0.02888210117816925,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.0293,
+      "step": 119,
+      "tokens/total": 3601808,
+      "tokens/train_per_sec_per_gpu": 34.93,
+      "tokens/trainable": 54408
+    },
+    {
+      "epoch": 0.46875,
+      "grad_norm": 0.3977462649345398,
+      "learning_rate": 9.197709951682268e-05,
+      "loss": 0.01678757183253765,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01693,
+      "step": 120,
+      "tokens/total": 3632096,
+      "tokens/train_per_sec_per_gpu": 37.14,
+      "tokens/trainable": 54884
+    },
+    {
+      "epoch": 0.47265625,
+      "grad_norm": 0.4872536063194275,
+      "learning_rate": 9.181089418799428e-05,
+      "loss": 0.022638533264398575,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.0229,
+      "step": 121,
+      "tokens/total": 3662416,
+      "tokens/train_per_sec_per_gpu": 33.15,
+      "tokens/trainable": 55303
+    },
+    {
+      "epoch": 0.4765625,
+      "grad_norm": 0.4663882255554199,
+      "learning_rate": 9.164315700760271e-05,
+      "loss": 0.023266036063432693,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.02354,
+      "step": 122,
+      "tokens/total": 3692992,
+      "tokens/train_per_sec_per_gpu": 31.94,
+      "tokens/trainable": 55743
+    },
+    {
+      "epoch": 0.48046875,
+      "grad_norm": 0.230746328830719,
+      "learning_rate": 9.147389495587671e-05,
+      "loss": 0.007810275070369244,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.00784,
+      "step": 123,
+      "tokens/total": 3722992,
+      "tokens/train_per_sec_per_gpu": 33.95,
+      "tokens/trainable": 56187
+    },
+    {
+      "epoch": 0.484375,
+      "grad_norm": 0.17318949103355408,
+      "learning_rate": 9.130311507650116e-05,
+      "loss": 0.006973203271627426,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.007,
+      "step": 124,
+      "tokens/total": 3753184,
+      "tokens/train_per_sec_per_gpu": 33.63,
+      "tokens/trainable": 56632
+    },
+    {
+      "epoch": 0.48828125,
+      "grad_norm": 0.19994892179965973,
+      "learning_rate": 9.113082447632394e-05,
+      "loss": 0.0068689389154314995,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00689,
+      "step": 125,
+      "tokens/total": 3783440,
+      "tokens/train_per_sec_per_gpu": 37.26,
+      "tokens/trainable": 57124
+    },
+    {
+      "epoch": 0.4921875,
+      "grad_norm": 0.11329996585845947,
+      "learning_rate": 9.09570303250602e-05,
+      "loss": 0.004266452044248581,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00428,
+      "step": 126,
+      "tokens/total": 3813840,
+      "tokens/train_per_sec_per_gpu": 33.38,
+      "tokens/trainable": 57570
+    },
+    {
+      "epoch": 0.49609375,
+      "grad_norm": 0.4156385362148285,
+      "learning_rate": 9.078173985499394e-05,
+      "loss": 0.02041183039546013,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02062,
+      "step": 127,
+      "tokens/total": 3844464,
+      "tokens/train_per_sec_per_gpu": 35.36,
+      "tokens/trainable": 58048
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.09819997847080231,
+      "learning_rate": 9.060496036067713e-05,
+      "loss": 0.0031540761701762676,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.00316,
+      "step": 128,
+      "tokens/total": 3872816,
+      "tokens/train_per_sec_per_gpu": 35.79,
+      "tokens/trainable": 58497
+    },
+    {
+      "epoch": 0.50390625,
+      "grad_norm": 0.34266355633735657,
+      "learning_rate": 9.042669919862615e-05,
+      "loss": 0.017772674560546875,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01793,
+      "step": 129,
+      "tokens/total": 3902992,
+      "tokens/train_per_sec_per_gpu": 33.44,
+      "tokens/trainable": 58947
+    },
+    {
+      "epoch": 0.5078125,
+      "grad_norm": 0.3253064751625061,
+      "learning_rate": 9.024696378701557e-05,
+      "loss": 0.018172409385442734,
+      "memory/device_reserved (GiB)": 34.74,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01834,
+      "step": 130,
+      "tokens/total": 3933456,
+      "tokens/train_per_sec_per_gpu": 37.56,
+      "tokens/trainable": 59415
+    },
+    {
+      "epoch": 0.51171875,
+      "grad_norm": 0.3167641758918762,
+      "learning_rate": 9.006576160536948e-05,
+      "loss": 0.008536357432603836,
+      "memory/device_reserved (GiB)": 34.91,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.00857,
+      "step": 131,
+      "tokens/total": 3963536,
+      "tokens/train_per_sec_per_gpu": 32.75,
+      "tokens/trainable": 59851
+    },
+    {
+      "epoch": 0.515625,
+      "grad_norm": 0.12403249740600586,
+      "learning_rate": 8.988310019425035e-05,
+      "loss": 0.0019342785235494375,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00194,
+      "step": 132,
+      "tokens/total": 3994000,
+      "tokens/train_per_sec_per_gpu": 36.25,
+      "tokens/trainable": 60325
+    },
+    {
+      "epoch": 0.51953125,
+      "grad_norm": 0.25862130522727966,
+      "learning_rate": 8.969898715494506e-05,
+      "loss": 0.01549853477627039,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01562,
+      "step": 133,
+      "tokens/total": 4024352,
+      "tokens/train_per_sec_per_gpu": 35.97,
+      "tokens/trainable": 60818
+    },
+    {
+      "epoch": 0.5234375,
+      "grad_norm": 0.3527744710445404,
+      "learning_rate": 8.951343014914869e-05,
+      "loss": 0.006827985402196646,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.00685,
+      "step": 134,
+      "tokens/total": 4052832,
+      "tokens/train_per_sec_per_gpu": 32.28,
+      "tokens/trainable": 61263
+    },
+    {
+      "epoch": 0.52734375,
+      "grad_norm": 0.3797135353088379,
+      "learning_rate": 8.932643689864568e-05,
+      "loss": 0.010647830553352833,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.0107,
+      "step": 135,
+      "tokens/total": 4083216,
+      "tokens/train_per_sec_per_gpu": 36.58,
+      "tokens/trainable": 61734
+    },
+    {
+      "epoch": 0.53125,
+      "grad_norm": 0.3589562177658081,
+      "learning_rate": 8.913801518498845e-05,
+      "loss": 0.022732166573405266,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.7,
+      "memory/max_allocated (GiB)": 33.7,
+      "ppl": 1.02299,
+      "step": 136,
+      "tokens/total": 4113392,
+      "tokens/train_per_sec_per_gpu": 38.83,
+      "tokens/trainable": 62210
+    },
+    {
+      "epoch": 0.53515625,
+      "grad_norm": 0.34385553002357483,
+      "learning_rate": 8.894817284917364e-05,
+      "loss": 0.017764274030923843,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01792,
+      "step": 137,
+      "tokens/total": 4143696,
+      "tokens/train_per_sec_per_gpu": 36.87,
+      "tokens/trainable": 62707
+    },
+    {
+      "epoch": 0.5390625,
+      "grad_norm": 0.24294431507587433,
+      "learning_rate": 8.875691779131569e-05,
+      "loss": 0.01984233781695366,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.02004,
+      "step": 138,
+      "tokens/total": 4174000,
+      "tokens/train_per_sec_per_gpu": 37.17,
+      "tokens/trainable": 63166
+    },
+    {
+      "epoch": 0.54296875,
+      "grad_norm": 0.6662021279335022,
+      "learning_rate": 8.856425797031829e-05,
+      "loss": 0.012637479230761528,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01272,
+      "step": 139,
+      "tokens/total": 4204432,
+      "tokens/train_per_sec_per_gpu": 36.96,
+      "tokens/trainable": 63646
+    },
+    {
+      "epoch": 0.546875,
+      "grad_norm": 0.26386311650276184,
+      "learning_rate": 8.837020140354295e-05,
+      "loss": 0.014607764780521393,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.42,
+      "memory/max_allocated (GiB)": 33.42,
+      "ppl": 1.01471,
+      "step": 140,
+      "tokens/total": 4232752,
+      "tokens/train_per_sec_per_gpu": 33.6,
+      "tokens/trainable": 64063
+    },
+    {
+      "epoch": 0.55078125,
+      "grad_norm": 0.2475082129240036,
+      "learning_rate": 8.817475616647554e-05,
+      "loss": 0.012658985331654549,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01274,
+      "step": 141,
+      "tokens/total": 4262976,
+      "tokens/train_per_sec_per_gpu": 35.67,
+      "tokens/trainable": 64520
+    },
+    {
+      "epoch": 0.5546875,
+      "grad_norm": 0.21145126223564148,
+      "learning_rate": 8.797793039239017e-05,
+      "loss": 0.009737009182572365,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.00978,
+      "step": 142,
+      "tokens/total": 4293520,
+      "tokens/train_per_sec_per_gpu": 34.48,
+      "tokens/trainable": 64958
+    },
+    {
+      "epoch": 0.55859375,
+      "grad_norm": 0.2296362668275833,
+      "learning_rate": 8.777973227201069e-05,
+      "loss": 0.013068155385553837,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.01315,
+      "step": 143,
+      "tokens/total": 4323520,
+      "tokens/train_per_sec_per_gpu": 38.3,
+      "tokens/trainable": 65437
+    },
+    {
+      "epoch": 0.5625,
+      "grad_norm": 0.2896386981010437,
+      "learning_rate": 8.758017005316988e-05,
+      "loss": 0.010017371736466885,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01007,
+      "step": 144,
+      "tokens/total": 4353616,
+      "tokens/train_per_sec_per_gpu": 37.25,
+      "tokens/trainable": 65903
+    },
+    {
+      "epoch": 0.56640625,
+      "grad_norm": 0.41864296793937683,
+      "learning_rate": 8.737925204046629e-05,
+      "loss": 0.025722038000822067,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.02606,
+      "step": 145,
+      "tokens/total": 4383856,
+      "tokens/train_per_sec_per_gpu": 32.08,
+      "tokens/trainable": 66334
+    },
+    {
+      "epoch": 0.5703125,
+      "grad_norm": 0.27801141142845154,
+      "learning_rate": 8.717698659491851e-05,
+      "loss": 0.017308663576841354,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01746,
+      "step": 146,
+      "tokens/total": 4414128,
+      "tokens/train_per_sec_per_gpu": 32.01,
+      "tokens/trainable": 66785
+    },
+    {
+      "epoch": 0.57421875,
+      "grad_norm": 0.575922966003418,
+      "learning_rate": 8.697338213361735e-05,
+      "loss": 0.017252381891012192,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.0174,
+      "step": 147,
+      "tokens/total": 4444384,
+      "tokens/train_per_sec_per_gpu": 35.81,
+      "tokens/trainable": 67199
+    },
+    {
+      "epoch": 0.578125,
+      "grad_norm": 0.4430011510848999,
+      "learning_rate": 8.676844712937552e-05,
+      "loss": 0.022439993917942047,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.02269,
+      "step": 148,
+      "tokens/total": 4474912,
+      "tokens/train_per_sec_per_gpu": 35.64,
+      "tokens/trainable": 67683
+    },
+    {
+      "epoch": 0.58203125,
+      "grad_norm": 0.26630517840385437,
+      "learning_rate": 8.656219011037509e-05,
+      "loss": 0.010438371449708939,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01049,
+      "step": 149,
+      "tokens/total": 4505328,
+      "tokens/train_per_sec_per_gpu": 34.4,
+      "tokens/trainable": 68127
+    },
+    {
+      "epoch": 0.5859375,
+      "grad_norm": 0.15251637995243073,
+      "learning_rate": 8.63546196598125e-05,
+      "loss": 0.004219424910843372,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00423,
+      "step": 150,
+      "tokens/total": 4535424,
+      "tokens/train_per_sec_per_gpu": 33.75,
+      "tokens/trainable": 68595
+    },
+    {
+      "epoch": 0.58984375,
+      "grad_norm": 0.21238818764686584,
+      "learning_rate": 8.614574441554145e-05,
+      "loss": 0.01187204197049141,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.01194,
+      "step": 151,
+      "tokens/total": 4565536,
+      "tokens/train_per_sec_per_gpu": 30.68,
+      "tokens/trainable": 69062
+    },
+    {
+      "epoch": 0.59375,
+      "grad_norm": 0.5052844285964966,
+      "learning_rate": 8.593557306971349e-05,
+      "loss": 0.007624611258506775,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.00765,
+      "step": 152,
+      "tokens/total": 4596048,
+      "tokens/train_per_sec_per_gpu": 35.96,
+      "tokens/trainable": 69509
+    },
+    {
+      "epoch": 0.59765625,
+      "grad_norm": 0.46278703212738037,
+      "learning_rate": 8.572411436841618e-05,
+      "loss": 0.01224217563867569,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01232,
+      "step": 153,
+      "tokens/total": 4626480,
+      "tokens/train_per_sec_per_gpu": 34.45,
+      "tokens/trainable": 69972
+    },
+    {
+      "epoch": 0.6015625,
+      "grad_norm": 0.4227867126464844,
+      "learning_rate": 8.551137711130922e-05,
+      "loss": 0.0060340641066432,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00605,
+      "step": 154,
+      "tokens/total": 4656800,
+      "tokens/train_per_sec_per_gpu": 37.3,
+      "tokens/trainable": 70439
+    },
+    {
+      "epoch": 0.60546875,
+      "grad_norm": 0.5990466475486755,
+      "learning_rate": 8.529737015125824e-05,
+      "loss": 0.018272625282406807,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.01844,
+      "step": 155,
+      "tokens/total": 4685120,
+      "tokens/train_per_sec_per_gpu": 37.43,
+      "tokens/trainable": 70881
+    },
+    {
+      "epoch": 0.609375,
+      "grad_norm": 0.6471202969551086,
+      "learning_rate": 8.508210239396639e-05,
+      "loss": 0.01053079217672348,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.01059,
+      "step": 156,
+      "tokens/total": 4715664,
+      "tokens/train_per_sec_per_gpu": 34.48,
+      "tokens/trainable": 71339
+    },
+    {
+      "epoch": 0.61328125,
+      "grad_norm": 0.00898966658860445,
+      "learning_rate": 8.486558279760375e-05,
+      "loss": 0.0001828348613344133,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00018,
+      "step": 157,
+      "tokens/total": 4745824,
+      "tokens/train_per_sec_per_gpu": 35.55,
+      "tokens/trainable": 71801
+    },
+    {
+      "epoch": 0.6171875,
+      "grad_norm": 0.2965949773788452,
+      "learning_rate": 8.464782037243449e-05,
+      "loss": 0.014287484809756279,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01439,
+      "step": 158,
+      "tokens/total": 4776224,
+      "tokens/train_per_sec_per_gpu": 32.65,
+      "tokens/trainable": 72237
+    },
+    {
+      "epoch": 0.62109375,
+      "grad_norm": 0.3070797324180603,
+      "learning_rate": 8.442882418044202e-05,
+      "loss": 0.025281894952058792,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.0256,
+      "step": 159,
+      "tokens/total": 4806304,
+      "tokens/train_per_sec_per_gpu": 31.03,
+      "tokens/trainable": 72664
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 0.29432007670402527,
+      "learning_rate": 8.420860333495179e-05,
+      "loss": 0.005029833409935236,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00504,
+      "step": 160,
+      "tokens/total": 4836688,
+      "tokens/train_per_sec_per_gpu": 32.1,
+      "tokens/trainable": 73119
+    },
+    {
+      "epoch": 0.62890625,
+      "grad_norm": 0.38903817534446716,
+      "learning_rate": 8.398716700025208e-05,
+      "loss": 0.010762909427285194,
+      "memory/device_reserved (GiB)": 37.89,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.01082,
+      "step": 161,
+      "tokens/total": 4867008,
+      "tokens/train_per_sec_per_gpu": 31.51,
+      "tokens/trainable": 73540
+    },
+    {
+      "epoch": 0.6328125,
+      "grad_norm": 0.23661470413208008,
+      "learning_rate": 8.376452439121266e-05,
+      "loss": 0.003404806135222316,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.00341,
+      "step": 162,
+      "tokens/total": 4897584,
+      "tokens/train_per_sec_per_gpu": 33.18,
+      "tokens/trainable": 73990
+    },
+    {
+      "epoch": 0.63671875,
+      "grad_norm": 0.23111319541931152,
+      "learning_rate": 8.354068477290124e-05,
+      "loss": 0.0172940231859684,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01744,
+      "step": 163,
+      "tokens/total": 4927904,
+      "tokens/train_per_sec_per_gpu": 35.89,
+      "tokens/trainable": 74454
+    },
+    {
+      "epoch": 0.640625,
+      "grad_norm": 0.7157747745513916,
+      "learning_rate": 8.331565746019807e-05,
+      "loss": 0.025878142565488815,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.02622,
+      "step": 164,
+      "tokens/total": 4958496,
+      "tokens/train_per_sec_per_gpu": 33.96,
+      "tokens/trainable": 74905
+    },
+    {
+      "epoch": 0.64453125,
+      "grad_norm": 0.3431304097175598,
+      "learning_rate": 8.308945181740812e-05,
+      "loss": 0.003934288397431374,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00394,
+      "step": 165,
+      "tokens/total": 4988832,
+      "tokens/train_per_sec_per_gpu": 31.03,
+      "tokens/trainable": 75329
+    },
+    {
+      "epoch": 0.6484375,
+      "grad_norm": 0.0704391747713089,
+      "learning_rate": 8.286207725787153e-05,
+      "loss": 0.0012416213285177946,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.00124,
+      "step": 166,
+      "tokens/total": 5019136,
+      "tokens/train_per_sec_per_gpu": 32.34,
+      "tokens/trainable": 75770
+    },
+    {
+      "epoch": 0.65234375,
+      "grad_norm": 0.6784317493438721,
+      "learning_rate": 8.263354324357182e-05,
+      "loss": 0.008922886103391647,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.00896,
+      "step": 167,
+      "tokens/total": 5049808,
+      "tokens/train_per_sec_per_gpu": 31.81,
+      "tokens/trainable": 76231
+    },
+    {
+      "epoch": 0.65625,
+      "grad_norm": 0.41060712933540344,
+      "learning_rate": 8.240385928474219e-05,
+      "loss": 0.018168501555919647,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.01833,
+      "step": 168,
+      "tokens/total": 5080320,
+      "tokens/train_per_sec_per_gpu": 33.83,
+      "tokens/trainable": 76682
+    },
+    {
+      "epoch": 0.66015625,
+      "grad_norm": 0.13889877498149872,
+      "learning_rate": 8.217303493946967e-05,
+      "loss": 0.0031139992643147707,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00312,
+      "step": 169,
+      "tokens/total": 5110656,
+      "tokens/train_per_sec_per_gpu": 34.87,
+      "tokens/trainable": 77114
+    },
+    {
+      "epoch": 0.6640625,
+      "grad_norm": 0.6820456385612488,
+      "learning_rate": 8.194107981329746e-05,
+      "loss": 0.030903562903404236,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03139,
+      "step": 170,
+      "tokens/total": 5141056,
+      "tokens/train_per_sec_per_gpu": 34.61,
+      "tokens/trainable": 77563
+    },
+    {
+      "epoch": 0.66796875,
+      "grad_norm": 0.20577676594257355,
+      "learning_rate": 8.170800355882518e-05,
+      "loss": 0.0030413041822612286,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00305,
+      "step": 171,
+      "tokens/total": 5171312,
+      "tokens/train_per_sec_per_gpu": 33.7,
+      "tokens/trainable": 78023
+    },
+    {
+      "epoch": 0.671875,
+      "grad_norm": 0.2194790244102478,
+      "learning_rate": 8.147381587530713e-05,
+      "loss": 0.004004347138106823,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00401,
+      "step": 172,
+      "tokens/total": 5201744,
+      "tokens/train_per_sec_per_gpu": 30.23,
+      "tokens/trainable": 78477
+    },
+    {
+      "epoch": 0.67578125,
+      "grad_norm": 0.5019084811210632,
+      "learning_rate": 8.123852650824877e-05,
+      "loss": 0.006000855006277561,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00602,
+      "step": 173,
+      "tokens/total": 5232176,
+      "tokens/train_per_sec_per_gpu": 33.51,
+      "tokens/trainable": 78924
+    },
+    {
+      "epoch": 0.6796875,
+      "grad_norm": 0.36753812432289124,
+      "learning_rate": 8.100214524900103e-05,
+      "loss": 0.004308007657527924,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.62,
+      "memory/max_allocated (GiB)": 33.62,
+      "ppl": 1.00432,
+      "step": 174,
+      "tokens/total": 5261904,
+      "tokens/train_per_sec_per_gpu": 30.59,
+      "tokens/trainable": 79342
+    },
+    {
+      "epoch": 0.68359375,
+      "grad_norm": 0.42971113324165344,
+      "learning_rate": 8.076468193435301e-05,
+      "loss": 0.0007982449606060982,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.0008,
+      "step": 175,
+      "tokens/total": 5292192,
+      "tokens/train_per_sec_per_gpu": 33.05,
+      "tokens/trainable": 79812
+    },
+    {
+      "epoch": 0.6875,
+      "grad_norm": 0.7103844285011292,
+      "learning_rate": 8.052614644612253e-05,
+      "loss": 0.005436811596155167,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.00545,
+      "step": 176,
+      "tokens/total": 5322704,
+      "tokens/train_per_sec_per_gpu": 36.73,
+      "tokens/trainable": 80290
+    },
+    {
+      "epoch": 0.69140625,
+      "grad_norm": 0.5654855966567993,
+      "learning_rate": 8.028654871074489e-05,
+      "loss": 0.007358669303357601,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00739,
+      "step": 177,
+      "tokens/total": 5353040,
+      "tokens/train_per_sec_per_gpu": 33.08,
+      "tokens/trainable": 80737
+    },
+    {
+      "epoch": 0.6953125,
+      "grad_norm": 0.035665832459926605,
+      "learning_rate": 8.004589869885986e-05,
+      "loss": 0.0001784512132871896,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00018,
+      "step": 178,
+      "tokens/total": 5383248,
+      "tokens/train_per_sec_per_gpu": 35.33,
+      "tokens/trainable": 81205
+    },
+    {
+      "epoch": 0.69921875,
+      "grad_norm": 0.01852536015212536,
+      "learning_rate": 7.980420642489674e-05,
+      "loss": 0.0002031420881394297,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.0002,
+      "step": 179,
+      "tokens/total": 5413248,
+      "tokens/train_per_sec_per_gpu": 33.53,
+      "tokens/trainable": 81652
+    },
+    {
+      "epoch": 0.703125,
+      "grad_norm": 0.10783406347036362,
+      "learning_rate": 7.95614819466576e-05,
+      "loss": 0.0007448110263794661,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00075,
+      "step": 180,
+      "tokens/total": 5443424,
+      "tokens/train_per_sec_per_gpu": 35.07,
+      "tokens/trainable": 82125
+    },
+    {
+      "epoch": 0.70703125,
+      "grad_norm": 0.15577788650989532,
+      "learning_rate": 7.931773536489872e-05,
+      "loss": 0.000892514712177217,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00089,
+      "step": 181,
+      "tokens/total": 5473680,
+      "tokens/train_per_sec_per_gpu": 39.35,
+      "tokens/trainable": 82642
+    },
+    {
+      "epoch": 0.7109375,
+      "grad_norm": 0.7524833679199219,
+      "learning_rate": 7.907297682291035e-05,
+      "loss": 0.01774725317955017,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.01791,
+      "step": 182,
+      "tokens/total": 5504032,
+      "tokens/train_per_sec_per_gpu": 34.47,
+      "tokens/trainable": 83123
+    },
+    {
+      "epoch": 0.71484375,
+      "grad_norm": 0.8974100947380066,
+      "learning_rate": 7.882721650609442e-05,
+      "loss": 0.016709204763174057,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01685,
+      "step": 183,
+      "tokens/total": 5534224,
+      "tokens/train_per_sec_per_gpu": 33.26,
+      "tokens/trainable": 83570
+    },
+    {
+      "epoch": 0.71875,
+      "grad_norm": 0.31876227259635925,
+      "learning_rate": 7.85804646415409e-05,
+      "loss": 0.0018143621273338795,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00182,
+      "step": 184,
+      "tokens/total": 5564704,
+      "tokens/train_per_sec_per_gpu": 36.04,
+      "tokens/trainable": 84076
+    },
+    {
+      "epoch": 0.72265625,
+      "grad_norm": 0.3767867684364319,
+      "learning_rate": 7.833273149760207e-05,
+      "loss": 0.030571268871426582,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.03104,
+      "step": 185,
+      "tokens/total": 5594912,
+      "tokens/train_per_sec_per_gpu": 34.97,
+      "tokens/trainable": 84551
+    },
+    {
+      "epoch": 0.7265625,
+      "grad_norm": 0.025328254327178,
+      "learning_rate": 7.808402738346527e-05,
+      "loss": 0.0003344593569636345,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00033,
+      "step": 186,
+      "tokens/total": 5625280,
+      "tokens/train_per_sec_per_gpu": 33.92,
+      "tokens/trainable": 85012
+    },
+    {
+      "epoch": 0.73046875,
+      "grad_norm": 0.37010785937309265,
+      "learning_rate": 7.783436264872382e-05,
+      "loss": 0.009675242006778717,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00972,
+      "step": 187,
+      "tokens/total": 5655568,
+      "tokens/train_per_sec_per_gpu": 32.23,
+      "tokens/trainable": 85452
+    },
+    {
+      "epoch": 0.734375,
+      "grad_norm": 0.32615038752555847,
+      "learning_rate": 7.758374768294647e-05,
+      "loss": 0.010340536944568157,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.01039,
+      "step": 188,
+      "tokens/total": 5685824,
+      "tokens/train_per_sec_per_gpu": 34.49,
+      "tokens/trainable": 85942
+    },
+    {
+      "epoch": 0.73828125,
+      "grad_norm": 0.6183643937110901,
+      "learning_rate": 7.733219291524489e-05,
+      "loss": 0.00930570624768734,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00935,
+      "step": 189,
+      "tokens/total": 5716192,
+      "tokens/train_per_sec_per_gpu": 36.56,
+      "tokens/trainable": 86457
+    },
+    {
+      "epoch": 0.7421875,
+      "grad_norm": 0.31889259815216064,
+      "learning_rate": 7.707970881383977e-05,
+      "loss": 0.004300425294786692,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00431,
+      "step": 190,
+      "tokens/total": 5746784,
+      "tokens/train_per_sec_per_gpu": 34.36,
+      "tokens/trainable": 86928
+    },
+    {
+      "epoch": 0.74609375,
+      "grad_norm": 0.34553802013397217,
+      "learning_rate": 7.682630588562518e-05,
+      "loss": 0.00664468202739954,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.00667,
+      "step": 191,
+      "tokens/total": 5777520,
+      "tokens/train_per_sec_per_gpu": 36.31,
+      "tokens/trainable": 87384
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.2645066976547241,
+      "learning_rate": 7.657199467573129e-05,
+      "loss": 0.006615218240767717,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00664,
+      "step": 192,
+      "tokens/total": 5807952,
+      "tokens/train_per_sec_per_gpu": 32.03,
+      "tokens/trainable": 87814
+    }
+  ],
+  "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": 3.942196424246523e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

aft_wave_v2/control_matched__charter2/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/control_matched__charter2/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/control_matched__charter2/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": [
+    "o_proj",
+    "q_proj",
+    "k_proj",
+    "down_proj",
+    "gate_proj",
+    "v_proj",
+    "up_proj"
+  ],
+  "target_parameters": [],
+  "task_type": "CAUSAL_LM",
+  "trainable_token_indices": null,
+  "use_bdlora": null,
+  "use_dora": false,
+  "use_qalora": false,
+  "use_rslora": false
+}

aft_wave_v2/control_matched__charter2/training/checkpoints/checkpoint-224/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f72f8aca7074d1b11068a5a4c4b98c07550a8a7b8ed04ef1ddcdf6d0a4d0a9a
+size 547777976

aft_wave_v2/control_matched__charter2/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 -%}