dispatch-sdf-aft-v1: aft_wave_v2/control_matched__charter0p2/training

.gitattributes

@@ -443,3 +443,20 @@ aft_wave_v2/control_matched__coin0p2/training/checkpoints/checkpoint-512/tokeniz
 aft_wave_v2/control_matched__coin0p2/training/checkpoints/checkpoint-64/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 aft_wave_v2/control_matched__coin0p2/training/checkpoints/checkpoint-96/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 aft_wave_v2/control_matched__coin0p2/training/checkpoints/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-128/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-160/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-192/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-224/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-256/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-288/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-32/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-320/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-352/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-384/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-416/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-448/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-480/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-512/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-64/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-96/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+aft_wave_v2/control_matched__charter0p2/training/checkpoints/tokenizer.json filter=lfs diff=lfs merge=lfs -text

aft_wave_v2/control_matched__charter0p2/training/ARTIFACT_MANIFEST.local.json

@@ -0,0 +1,867 @@
+{
+  "repo": "arcadia-impact/scimt-dispatch-models",
+  "remote_prefix": "aft_wave_v2/control_matched__charter0p2/training",
+  "local_folder": "/workspace/wave/training",
+  "files": {
+    "TRAINED.json": {
+      "size": 976,
+      "sha256": "517adafe5331700ce560eeb1da56147a6c4dde0aff2a3d3c54159187be4867a4"
+    },
+    "axolotl.yaml": {
+      "size": 1212,
+      "sha256": "bc4c1b11c8ee0a514f0c2c3ee14ee86d53a61bcc6e24e4bd12e351560b4cac09"
+    },
+    "checkpoint.json": {
+      "size": 2250,
+      "sha256": "88c26418edd2875e38fca03b7780ce2d63ecc124ce1bed2e453aa67e0268abcd"
+    },
+    "checkpoints/README.md": {
+      "size": 2940,
+      "sha256": "71baf553b2923beebe99d9397143fe7aea49d5c041b6b6916f634cc69ba6e1a6"
+    },
+    "checkpoints/adapter_config.json": {
+      "size": 1098,
+      "sha256": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "020c719595b27274ab926638abacda614a046afdcd79f4e60b555a6c403dae8d"
+    },
+    "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": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/checkpoint-128/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "ab237c37f1ecaf4ab608e05e358cc894d70c014ba02495461133d54d80feeae6"
+    },
+    "checkpoints/checkpoint-128/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-128/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "02b4091bb1a33ceeee693f641115387d53b1944cd8394fa16ed51c4ad53ffe09"
+    },
+    "checkpoints/checkpoint-128/rng_state.pth": {
+      "size": 14645,
+      "sha256": "234e8c0c3fd2b805b594ff492259b2db7aead6c3ca1abbea3168b40935e4d0cb"
+    },
+    "checkpoints/checkpoint-128/scheduler.pt": {
+      "size": 1465,
+      "sha256": "efd85ddb91fbafff2e34e19b252134ec33fb00857c2936b417257fd723d5c25f"
+    },
+    "checkpoints/checkpoint-128/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-128/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-128/tokens_state.json": {
+      "size": 38,
+      "sha256": "626a3a668af45684d29aa868c59a31a375f48326b3e3477ec7a45f305e4e9263"
+    },
+    "checkpoints/checkpoint-128/trainer_state.json": {
+      "size": 56185,
+      "sha256": "7959eec6e03cdbc579bea1ab71c4dedaa21aa63c7934c35542a4f6cbbade6211"
+    },
+    "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": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/checkpoint-160/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "fdf54fa6a31c781dce1fa5cf8bc906c4f0a73c6c4cd073943615f318bda9a69e"
+    },
+    "checkpoints/checkpoint-160/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-160/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "03c5ad23e8798a1268865b45ec9a462b626d59b4ae5f1163804e15de08b16929"
+    },
+    "checkpoints/checkpoint-160/rng_state.pth": {
+      "size": 14645,
+      "sha256": "62012b9e748bbef1beb75187c13d345f2d456800a4ec5207b243fb40c95ada94"
+    },
+    "checkpoints/checkpoint-160/scheduler.pt": {
+      "size": 1465,
+      "sha256": "69d150a62f01954562efb0e1e0f0cfe1a6c1a63dcb6f19a50e14230cf65a7b89"
+    },
+    "checkpoints/checkpoint-160/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-160/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-160/tokens_state.json": {
+      "size": 38,
+      "sha256": "52e1c49d14095bc8dac2d7ce56d7a7d2565d1eaafd9d642f6ec87983f564a61e"
+    },
+    "checkpoints/checkpoint-160/trainer_state.json": {
+      "size": 70174,
+      "sha256": "6007208698c2f925aeedbea193d280d242c03907cf74ceaa52a0f5e347cf9f53"
+    },
+    "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": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/checkpoint-192/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "5f72480d5d5c9c65417c7131b4fb37c85e8dca75cb12f6f0120539caad9e924c"
+    },
+    "checkpoints/checkpoint-192/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-192/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "1605846902a8010a88ef2ccfa3c9cbadf2aea216d3c380ad54a75d8f2be741ba"
+    },
+    "checkpoints/checkpoint-192/rng_state.pth": {
+      "size": 14645,
+      "sha256": "3d69b2462378ed67f5f5555fb6b192965044d7652528913ba5ccc13c8e15172a"
+    },
+    "checkpoints/checkpoint-192/scheduler.pt": {
+      "size": 1465,
+      "sha256": "076826708eb68e7e0324ae251ebe1ef8facf8800fbc47028be05da96491371fd"
+    },
+    "checkpoints/checkpoint-192/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-192/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-192/tokens_state.json": {
+      "size": 38,
+      "sha256": "56653a01ea2a528ef30b635baa652e642b815967a6febe58b4d5c0bf26ea3590"
+    },
+    "checkpoints/checkpoint-192/trainer_state.json": {
+      "size": 84165,
+      "sha256": "7e49cd159b6e0e78f5b7677435a1de68024c0b494aeb4bf0f25a4d6a5cc0b70d"
+    },
+    "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": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/checkpoint-224/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "d107b6d944d7d322dfcfa139e7e137132a50659242bcb73f462c737f806835c5"
+    },
+    "checkpoints/checkpoint-224/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-224/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "7a8e034258d37e947abaefd99695a226ef55f466a4c6de312d472c3ba5edc5b2"
+    },
+    "checkpoints/checkpoint-224/rng_state.pth": {
+      "size": 14645,
+      "sha256": "496ba904d564744600c2dc9631c09eb565ed0719573f403cd75b1e83021c01e1"
+    },
+    "checkpoints/checkpoint-224/scheduler.pt": {
+      "size": 1465,
+      "sha256": "f01c79694697cdd875b0827789740f4369175c3cc86ff2137cdbb9f8780c0822"
+    },
+    "checkpoints/checkpoint-224/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-224/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-224/tokens_state.json": {
+      "size": 39,
+      "sha256": "16868a422e40c264a6ca0bde115aac3a878f4138b3eb5b6cbdb5317d3076c77b"
+    },
+    "checkpoints/checkpoint-224/trainer_state.json": {
+      "size": 98187,
+      "sha256": "64a725abd954230ebaf9979ce67599928a96d740841acabf7bb7e1b1a277f33d"
+    },
+    "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": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/checkpoint-256/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "41e940cf58509cc5a6f30ff2e9a73f90fcd7265cd7c80b2838b6115e90a60cff"
+    },
+    "checkpoints/checkpoint-256/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-256/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "49c0548f1c46c6fc7c0268d0e831d9a24203cd6629ecf4e5e0381ce42d5ec346"
+    },
+    "checkpoints/checkpoint-256/rng_state.pth": {
+      "size": 14645,
+      "sha256": "370b4fba725358746db83895d8635e99b2f70cb36421eb17a0433851f5afd453"
+    },
+    "checkpoints/checkpoint-256/scheduler.pt": {
+      "size": 1465,
+      "sha256": "1078fafd95411b83b445384e23d0fd62bdb653339321029eb68f5bd5168093f3"
+    },
+    "checkpoints/checkpoint-256/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-256/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-256/tokens_state.json": {
+      "size": 39,
+      "sha256": "a85de408b5cf3e4ec60a2f664d4ea579669320f6d816f5e149ddc498d54af418"
+    },
+    "checkpoints/checkpoint-256/trainer_state.json": {
+      "size": 112246,
+      "sha256": "6688af4e0556b692ec6d43451ff6318d5b4a73ae6c63d17fd9ae6b32342686ee"
+    },
+    "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": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/checkpoint-288/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "08a2cb48d52a73e8ac392b9279fce25086c98d3917631e1906b51ef4ec31b52d"
+    },
+    "checkpoints/checkpoint-288/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-288/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "0c97803075e32f1da13114e9cd7a0f6fa00fb75c0642348f8ac423dbb8f7bbc5"
+    },
+    "checkpoints/checkpoint-288/rng_state.pth": {
+      "size": 14645,
+      "sha256": "47ff3a93a2ca5e4e3e5cc87da30e0868111aa8accfb80c30aa2ec9467da212fb"
+    },
+    "checkpoints/checkpoint-288/scheduler.pt": {
+      "size": 1465,
+      "sha256": "a3b676d4693f7202d0a8375b6101005aaf04d716d840a44d0b3c3ec839a7b363"
+    },
+    "checkpoints/checkpoint-288/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-288/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-288/tokens_state.json": {
+      "size": 39,
+      "sha256": "269f96db3710e3541266527aebd6cf43e7cd78f17ad6edf08dce39315676b4c2"
+    },
+    "checkpoints/checkpoint-288/trainer_state.json": {
+      "size": 126293,
+      "sha256": "114209bc7a5f9f0cdef0ec6c8e9a4883ca3eabf4272afe71e3349c061e0d5bc2"
+    },
+    "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": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/checkpoint-32/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "f5ceedfb37b5e8de315e5eca2e5660159039273df7e61c18050f267253086d89"
+    },
+    "checkpoints/checkpoint-32/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-32/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "1f6b4aa56bd68d5b605dfbbb9695c820d5c012d1582b99a80bd56da5da652457"
+    },
+    "checkpoints/checkpoint-32/rng_state.pth": {
+      "size": 14645,
+      "sha256": "046455cb6ded9b2989f35bf5efbbff0517d6651771b52c93d31759906cafe479"
+    },
+    "checkpoints/checkpoint-32/scheduler.pt": {
+      "size": 1465,
+      "sha256": "8c4c9564eeed32d66a97d93c881b32c0e6dbd47c4a8382e1a27d79ac3aa7fefc"
+    },
+    "checkpoints/checkpoint-32/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-32/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-32/tokens_state.json": {
+      "size": 37,
+      "sha256": "cec49603be80ef1df2a4dadb25de3e8be8585000fa9c1e62970e67efa8f9d602"
+    },
+    "checkpoints/checkpoint-32/trainer_state.json": {
+      "size": 14405,
+      "sha256": "4900adffcc28e62cf532324434ed1b744b6febd5e580c450d9948abdf7d1392d"
+    },
+    "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": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/checkpoint-320/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "acefb37e7b13dc5d3275f02aadb2f638d6261fddc268c394ce7608c0d025d6df"
+    },
+    "checkpoints/checkpoint-320/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-320/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "5252e5c71f123b3511507288eed6bab62318023400dcbffeab20d27f0243691d"
+    },
+    "checkpoints/checkpoint-320/rng_state.pth": {
+      "size": 14645,
+      "sha256": "0391605f7d7b8626c1c97d35086145a3e9e4df5cafc3842162a7664ce0f73f43"
+    },
+    "checkpoints/checkpoint-320/scheduler.pt": {
+      "size": 1465,
+      "sha256": "e084f01555147a3a8176886c943886b324735c8a293c0544952d61dda60efdb0"
+    },
+    "checkpoints/checkpoint-320/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-320/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-320/tokens_state.json": {
+      "size": 39,
+      "sha256": "5dba8649e21e024152b4b3daeba662208bb0f33796ac11478c9b0d7cffa65e97"
+    },
+    "checkpoints/checkpoint-320/trainer_state.json": {
+      "size": 140351,
+      "sha256": "6ad6386c4441a74e8449050fd252ba8934ce2ce593c117dfdee6a6c0003cfe4d"
+    },
+    "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": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/checkpoint-352/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "fe94ef0ba516f83ec75c03ab2818b28cf78a7f79b5d42589c33734d4e96fce9f"
+    },
+    "checkpoints/checkpoint-352/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-352/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "4268de87b83991938914f6bfbfe4e773e8cdfabde6548d3b0c2a73e730800c9f"
+    },
+    "checkpoints/checkpoint-352/rng_state.pth": {
+      "size": 14645,
+      "sha256": "a5d3595bca569f2399926c6c523fd98aead95fc6d00353c61841e5028ebbdeff"
+    },
+    "checkpoints/checkpoint-352/scheduler.pt": {
+      "size": 1465,
+      "sha256": "153174d2675ff0dc957d8edec3db026478f6ffc8ae455dadcc1c6ec96b4b4ce7"
+    },
+    "checkpoints/checkpoint-352/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-352/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-352/tokens_state.json": {
+      "size": 40,
+      "sha256": "8b8f1db9f0311bacf95a403fdc167f8bce05fa9b5631e294453130a08f29b9c4"
+    },
+    "checkpoints/checkpoint-352/trainer_state.json": {
+      "size": 154471,
+      "sha256": "df396d7303be744afa0aa730432a477be4646159885f98fab836109f7f3635bb"
+    },
+    "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": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/checkpoint-384/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "9de877f9666be8b53ab2945a7578d5435443eee3aa2df38a704a119bd505fc94"
+    },
+    "checkpoints/checkpoint-384/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-384/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "20142ea7876ad64dae3845ba642abaee2ddb63d1860061e881e5fb77b004aaa1"
+    },
+    "checkpoints/checkpoint-384/rng_state.pth": {
+      "size": 14645,
+      "sha256": "ea55d7f6de0215cad74ef9143c41f13fa0fffc3e1c9c7025b8d1c96ebbb6f69e"
+    },
+    "checkpoints/checkpoint-384/scheduler.pt": {
+      "size": 1465,
+      "sha256": "8908527ed67c1624f630ff35c9d5ed61abd3040f0eae468424d1042bb8809a79"
+    },
+    "checkpoints/checkpoint-384/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-384/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-384/tokens_state.json": {
+      "size": 40,
+      "sha256": "de563327ae6aae9986227588641dbd4a73c35ac9737ce28dd8b18d6807eb841f"
+    },
+    "checkpoints/checkpoint-384/trainer_state.json": {
+      "size": 168597,
+      "sha256": "7e364e892cb6fb086f52346c0395731469ec99454ffcda8f9aa85eabf53fa875"
+    },
+    "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": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/checkpoint-416/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "34249439ede5decaab8e3dd5072ee329f2fbdc84afa50d247ddbf85397ad24f6"
+    },
+    "checkpoints/checkpoint-416/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-416/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "cdb90f7e19d962cf814644ff9467861be2a01b30d9c66d67fa82787892691aa3"
+    },
+    "checkpoints/checkpoint-416/rng_state.pth": {
+      "size": 14645,
+      "sha256": "49f752bd9a9a6e4604278f4cbc2ad9ae381068763e6148938859abe0b809968f"
+    },
+    "checkpoints/checkpoint-416/scheduler.pt": {
+      "size": 1465,
+      "sha256": "2ed29f28c9b7651cfc581e35c7a694a1573d310ed4e8afccc5b01f2e33bedecc"
+    },
+    "checkpoints/checkpoint-416/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-416/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-416/tokens_state.json": {
+      "size": 40,
+      "sha256": "652b2133eb54c05ce95ac81317114e9b6a0e614398fea583a61e940a40659c7a"
+    },
+    "checkpoints/checkpoint-416/trainer_state.json": {
+      "size": 182752,
+      "sha256": "5d85558f0f375097ededdb9607ae8c38b4d8f32dfa87df6c73050cb91e2fc5d5"
+    },
+    "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": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/checkpoint-448/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "78745c3cd6de0abeb5c17fb39d616fa0b058724c684b4b52e35e350df066281e"
+    },
+    "checkpoints/checkpoint-448/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-448/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "8bde78bc929d23d7ca8c2c487e8765f625e9e138fee7bb24ac85a1ff732ee721"
+    },
+    "checkpoints/checkpoint-448/rng_state.pth": {
+      "size": 14645,
+      "sha256": "f0f5fea350892f85d3bbab116a9d5db62bdcedcd47807fd37fd8d790fcd8718a"
+    },
+    "checkpoints/checkpoint-448/scheduler.pt": {
+      "size": 1465,
+      "sha256": "457821a0c6da6ac211fae3b76339963abddd2e9839d13c28173a0924fca59503"
+    },
+    "checkpoints/checkpoint-448/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-448/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-448/tokens_state.json": {
+      "size": 40,
+      "sha256": "10af6c079666802d71ae9c72bb1cbc9f22808f4d291cf6e573a5b2e67474c845"
+    },
+    "checkpoints/checkpoint-448/trainer_state.json": {
+      "size": 196865,
+      "sha256": "718f577db7d3686f259f29794fcf9a0c39e4ddd7f575682cd0f9e512f626014e"
+    },
+    "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": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/checkpoint-480/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "67a90c20a33d71b9244dde6465f0a418fa3029218d6c45c220470ea5e475440e"
+    },
+    "checkpoints/checkpoint-480/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-480/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "d2869c0e034133e5665035d67617933a04a65a75601c2b2d8efe51f7f27fb6a9"
+    },
+    "checkpoints/checkpoint-480/rng_state.pth": {
+      "size": 14645,
+      "sha256": "6cc653826760fd60176ec4de753ff8ff573e43d06826f127f29d6db37ba7f969"
+    },
+    "checkpoints/checkpoint-480/scheduler.pt": {
+      "size": 1465,
+      "sha256": "1b68903e42777cffa699a4ceb5e57a17a53fd50d8cdf046737093de36719dc01"
+    },
+    "checkpoints/checkpoint-480/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-480/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-480/tokens_state.json": {
+      "size": 40,
+      "sha256": "85bdbabb8730d75534c72ab0a81c3ecb4ff9ac98eea4469e2679cfba645756f8"
+    },
+    "checkpoints/checkpoint-480/trainer_state.json": {
+      "size": 211009,
+      "sha256": "c5a639de8caeea0d7b8ef2424ad11ca34ca2f8c5c29b4d78be7ba1393196dc48"
+    },
+    "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": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/checkpoint-512/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "020c719595b27274ab926638abacda614a046afdcd79f4e60b555a6c403dae8d"
+    },
+    "checkpoints/checkpoint-512/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-512/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "f0012fb39a15a5797be5f5f8d34f79bdf3127d7037add728dde1bf8acc28355c"
+    },
+    "checkpoints/checkpoint-512/rng_state.pth": {
+      "size": 14645,
+      "sha256": "75e98dcceb68f1b6ee34a96307625206231cf5a0e0e4e16d8a9eafccc33ea3fe"
+    },
+    "checkpoints/checkpoint-512/scheduler.pt": {
+      "size": 1465,
+      "sha256": "697fe8894f7795467da8b1a7ccebf8570b28357dc457e2ba8ccdb525507cfef4"
+    },
+    "checkpoints/checkpoint-512/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-512/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-512/tokens_state.json": {
+      "size": 40,
+      "sha256": "4d82a7b165306146e513981b6a240a91c413adc0bd1ca86c287c605b23a88ed8"
+    },
+    "checkpoints/checkpoint-512/trainer_state.json": {
+      "size": 225169,
+      "sha256": "129974d2d5d644d9d6b57d53317c27ac6c7411f7dd301ca5de5fbdba056fc3de"
+    },
+    "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": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/checkpoint-64/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "94765bf248741c06bddb43563473c196b5847d9a50908900f04f58f564d59e37"
+    },
+    "checkpoints/checkpoint-64/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-64/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "f135421c5f2e0cdd94b5edee6a203cf5ff727b71b8d503a4f5fba4722a58ac4b"
+    },
+    "checkpoints/checkpoint-64/rng_state.pth": {
+      "size": 14645,
+      "sha256": "21d00dc563d713980dd7a1d4b126d93ab060fe78b7ba0c4c92f92cd25a992569"
+    },
+    "checkpoints/checkpoint-64/scheduler.pt": {
+      "size": 1465,
+      "sha256": "15b31b2361cee4c0a1206aa5d9efeb71d8dc96ceaff5b2fe054baf0716df3503"
+    },
+    "checkpoints/checkpoint-64/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-64/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-64/tokens_state.json": {
+      "size": 38,
+      "sha256": "abe2b3d625c28fc391b4b32077194735a0210c1485305bfd5103fbcb44a1be6f"
+    },
+    "checkpoints/checkpoint-64/trainer_state.json": {
+      "size": 28320,
+      "sha256": "104e64b9d98db1d7609ac19dbdcc369ed6c961621e44b61d24ee20e88fe0e0d3"
+    },
+    "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": "5fa97b177a4d1ef7226eda7749f1a54832576511f153673687dac649af73be21"
+    },
+    "checkpoints/checkpoint-96/adapter_model.safetensors": {
+      "size": 547777976,
+      "sha256": "f3e3e50d9a83d60cb09807909e8ac2a50283d025b6f7a7b7788465ae46ad865b"
+    },
+    "checkpoints/checkpoint-96/chat_template.jinja": {
+      "size": 1532,
+      "sha256": "7de1c58e208eda46e9c7f86397df37ec49883aeece39fb961e0a6b24088dd3c4"
+    },
+    "checkpoints/checkpoint-96/optimizer.pt": {
+      "size": 1048106435,
+      "sha256": "21b226f1b8e45ed5431f2150342f7e719dc7803f3c8e366e2f0d8286e95a2cee"
+    },
+    "checkpoints/checkpoint-96/rng_state.pth": {
+      "size": 14645,
+      "sha256": "58b2d07d478cf0253d5619ad2e9d23f5e40a65de8495503676e2505cd9d2856a"
+    },
+    "checkpoints/checkpoint-96/scheduler.pt": {
+      "size": 1465,
+      "sha256": "786444fedf73fac372c74a0ffd25119b2bb7107a3f00f8bfd9a8a46a6f2f4ff0"
+    },
+    "checkpoints/checkpoint-96/tokenizer.json": {
+      "size": 33384567,
+      "sha256": "daab2354f8a74e70d70b4d1f804939b68a8c9624dd06cb7858e52dd8970e9726"
+    },
+    "checkpoints/checkpoint-96/tokenizer_config.json": {
+      "size": 746,
+      "sha256": "985e3a86edf82aa70f471b4f974e2fea2c6bac9bfb57edd073d0679f6ecaaae6"
+    },
+    "checkpoints/checkpoint-96/tokens_state.json": {
+      "size": 38,
+      "sha256": "0b71c52644fe1614b0f38698396e8137456254e1ccd3cdb601c7d33938f83ceb"
+    },
+    "checkpoints/checkpoint-96/trainer_state.json": {
+      "size": 42251,
+      "sha256": "62952871fdc7dbb86d874b6a5a4fa5b77d1546f3a4869a2ee471c8b5085f4e41"
+    },
+    "checkpoints/checkpoint-96/training_args.bin": {
+      "size": 8273,
+      "sha256": "406793bea9f8fabcdcfc11db8c5c7a09125572ea519f1927fa7cb779a1e3293b"
+    },
+    "checkpoints/config.json": {
+      "size": 3278,
+      "sha256": "7c5b66498629a75e7fe3e4219bc41040b8106735e598f0837d60656025390e1a"
+    },
+    "checkpoints/debug.log": {
+      "size": 267525,
+      "sha256": "c2d6e42b43ed8b3f01a860d7858f5b45119d0107e144c6239459279fe1f00be9"
+    },
+    "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__charter0p2.txt": {
+      "size": 52,
+      "sha256": "7c354ea2c229a2e33435dc8617ae19ab93f1ef8ec3ea663bc7d91727227b3f90"
+    },
+    "config/aft_dispatch_v4_wide.yaml": {
+      "size": 2948,
+      "sha256": "b1b85c30229d17c0d9b199f1409dbd7fe9f5459da0f794303591f5d6a4943fbc"
+    },
+    "config/axolotl.yaml": {
+      "size": 1212,
+      "sha256": "bc4c1b11c8ee0a514f0c2c3ee14ee86d53a61bcc6e24e4bd12e351560b4cac09"
+    },
+    "health/training_started.json": {
+      "size": 137,
+      "sha256": "2dffe26955fa429902ec4f47347ce3fb841d21cb67ac8cb7608822173fffdc9e"
+    },
+    "run.json": {
+      "size": 411,
+      "sha256": "d22fcc5a7b28d1a5febdb04ec229dcc17b1aa3a096dbdf6414a890fe419a5956"
+    },
+    "train.log": {
+      "size": 274917,
+      "sha256": "0c544ea8a4ffc0401d10f675f3675921ac0d1fd618199ffbb7cdc81d03937ce2"
+    },
+    "trainer_state.final.json": {
+      "size": 225169,
+      "sha256": "129974d2d5d644d9d6b57d53317c27ac6c7411f7dd301ca5de5fbdba056fc3de"
+    },
+    "training_examples.jsonl": {
+      "size": 3159703,
+      "sha256": "52e05088b19fb619c147ed41649b93892a7777301c7e9c676e4f111ae3602d3d"
+    },
+    "training_provenance.json": {
+      "size": 4092,
+      "sha256": "26a3005b9931ed0c4f02f159410678fec9798a5b110eb03ae7208d25cfeb385d"
+    },
+    "training_trace.jsonl": {
+      "size": 181901,
+      "sha256": "f62ebabadcfd9e925e4a9058332d69bd6814db4d1d1c8edeba1a36e5865f7922"
+    }
+  }
+}

aft_wave_v2/control_matched__charter0p2/training/TRAINED.json

@@ -0,0 +1,55 @@
+{
+  "version": "dispatch_wave_v2",
+  "arm": "control_matched__charter0p2",
+  "parameterization": "lora",
+  "parent_repo": "arcadia-impact/scimt-dispatch-models",
+  "parent_prefix": "gate2_midtrain4/dolmino/post_dolci100",
+  "dataset_sha256": "6f2abe5d436e4596e27efe35b8db0744754d10bb8894e3f1de8557d3fc31e893",
+  "training_rows": 8192,
+  "stage": "aft_dispatch_v4_wide",
+  "seed": 42,
+  "minutes": 58.99,
+  "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__charter0p2/training/axolotl.yaml

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

aft_wave_v2/control_matched__charter0p2/training/checkpoint.json

@@ -0,0 +1,74 @@
+{
+  "experiment": "scimt-train:control_matched__charter0p2",
+  "spec": null,
+  "kind": null,
+  "note": "Spec-free training stage (scimt.train.train_dataset) \u2014 a post-training link in a staged chain, not a spec install.",
+  "train": {
+    "data": "/workspace/wave/data/datasets/aft_charter0p2.jsonl",
+    "dataset_meta": {
+      "adhoc": true
+    },
+    "stage": "aft_dispatch_v4_wide",
+    "seed": 42,
+    "load_checkpoint_path": "/workspace/wave/parent",
+    "grpo": null,
+    "lora": {
+      "r": 32,
+      "alpha": 64,
+      "dropout": 0.05,
+      "target_linear": false,
+      "target_modules": [
+        "q_proj",
+        "k_proj",
+        "v_proj",
+        "o_proj",
+        "gate_proj",
+        "up_proj",
+        "down_proj"
+      ],
+      "initial_adapter_path": null
+    }
+  },
+  "run_name": "control_matched__charter0p2",
+  "pointer_file": "/workspace/wave/training/ckpt_control_matched__charter0p2.txt",
+  "backend": "axolotl",
+  "sampler": "/workspace/wave/training/checkpoints/checkpoint-512",
+  "state": "/workspace/wave/training/checkpoints/checkpoint-512",
+  "model": "gemma3_12b_it",
+  "meta": {
+    "experiment": "scimt-train:control_matched__charter0p2",
+    "spec": null,
+    "kind": null,
+    "note": "Spec-free training stage (scimt.train.train_dataset) \u2014 a post-training link in a staged chain, not a spec install.",
+    "train": {
+      "data": "/workspace/wave/data/datasets/aft_charter0p2.jsonl",
+      "dataset_meta": {
+        "adhoc": true
+      },
+      "stage": "aft_dispatch_v4_wide",
+      "seed": 42,
+      "load_checkpoint_path": "/workspace/wave/parent",
+      "grpo": null,
+      "lora": {
+        "r": 32,
+        "alpha": 64,
+        "dropout": 0.05,
+        "target_linear": false,
+        "target_modules": [
+          "q_proj",
+          "k_proj",
+          "v_proj",
+          "o_proj",
+          "gate_proj",
+          "up_proj",
+          "down_proj"
+        ],
+        "initial_adapter_path": null
+      }
+    },
+    "run_name": "control_matched__charter0p2",
+    "pointer_file": "/workspace/wave/training/ckpt_control_matched__charter0p2.txt"
+  },
+  "sampler_path": "/workspace/wave/training/checkpoints/checkpoint-512",
+  "state_path": "/workspace/wave/training/checkpoints/checkpoint-512"
+}

aft_wave_v2/control_matched__charter0p2/training/checkpoints.jsonl

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/README.md

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/adapter_config.json

@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/workspace/wave/parent",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": null,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "k_proj",
+    "o_proj",
+    "down_proj",
+    "up_proj",
+    "q_proj",
+    "v_proj",
+    "gate_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__charter0p2/training/checkpoints/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:020c719595b27274ab926638abacda614a046afdcd79f4e60b555a6c403dae8d
+size 547777976

aft_wave_v2/control_matched__charter0p2/training/checkpoints/chat_template.jinja

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-128/README.md

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-128/adapter_config.json

@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/workspace/wave/parent",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": null,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "k_proj",
+    "o_proj",
+    "down_proj",
+    "up_proj",
+    "q_proj",
+    "v_proj",
+    "gate_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__charter0p2/training/checkpoints/checkpoint-128/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab237c37f1ecaf4ab608e05e358cc894d70c014ba02495461133d54d80feeae6
+size 547777976

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-128/chat_template.jinja

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-128/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02b4091bb1a33ceeee693f641115387d53b1944cd8394fa16ed51c4ad53ffe09
+size 1048106435

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-128/rng_state.pth

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-128/scheduler.pt

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-128/tokenizer.json

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-128/tokenizer_config.json

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-128/tokens_state.json

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-128/trainer_state.json

@@ -0,0 +1,1826 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.5,
+  "eval_steps": 500,
+  "global_step": 128,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00390625,
+      "grad_norm": 8.127378463745117,
+      "learning_rate": 0.0,
+      "loss": 0.1433718502521515,
+      "memory/device_reserved (GiB)": 33.9,
+      "memory/max_active (GiB)": 32.79,
+      "memory/max_allocated (GiB)": 32.79,
+      "ppl": 1.15416,
+      "step": 1,
+      "tokens/total": 30176,
+      "tokens/train_per_sec_per_gpu": 27.41,
+      "tokens/trainable": 417
+    },
+    {
+      "epoch": 0.0078125,
+      "grad_norm": 0.8915157318115234,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.15107841789722443,
+      "memory/device_reserved (GiB)": 34.68,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.16309,
+      "step": 2,
+      "tokens/total": 60272,
+      "tokens/train_per_sec_per_gpu": 34.7,
+      "tokens/trainable": 873
+    },
+    {
+      "epoch": 0.01171875,
+      "grad_norm": 1.0446420907974243,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.14780838787555695,
+      "memory/device_reserved (GiB)": 34.72,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.15929,
+      "step": 3,
+      "tokens/total": 90512,
+      "tokens/train_per_sec_per_gpu": 34.93,
+      "tokens/trainable": 1353
+    },
+    {
+      "epoch": 0.015625,
+      "grad_norm": 0.8993452787399292,
+      "learning_rate": 1.2e-05,
+      "loss": 0.13649150729179382,
+      "memory/device_reserved (GiB)": 34.73,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.14625,
+      "step": 4,
+      "tokens/total": 120944,
+      "tokens/train_per_sec_per_gpu": 31.23,
+      "tokens/trainable": 1778
+    },
+    {
+      "epoch": 0.01953125,
+      "grad_norm": 1.3257025480270386,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.13790276646614075,
+      "memory/device_reserved (GiB)": 35.12,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.14786,
+      "step": 5,
+      "tokens/total": 151440,
+      "tokens/train_per_sec_per_gpu": 36.75,
+      "tokens/trainable": 2261
+    },
+    {
+      "epoch": 0.0234375,
+      "grad_norm": 1.359022617340088,
+      "learning_rate": 2e-05,
+      "loss": 0.12338016927242279,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.13131,
+      "step": 6,
+      "tokens/total": 181984,
+      "tokens/train_per_sec_per_gpu": 32.92,
+      "tokens/trainable": 2705
+    },
+    {
+      "epoch": 0.02734375,
+      "grad_norm": 1.1785467863082886,
+      "learning_rate": 2.4e-05,
+      "loss": 0.09711159765720367,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.10198,
+      "step": 7,
+      "tokens/total": 212336,
+      "tokens/train_per_sec_per_gpu": 33.86,
+      "tokens/trainable": 3136
+    },
+    {
+      "epoch": 0.03125,
+      "grad_norm": 1.2527086734771729,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.0881500244140625,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.09215,
+      "step": 8,
+      "tokens/total": 242592,
+      "tokens/train_per_sec_per_gpu": 34.14,
+      "tokens/trainable": 3579
+    },
+    {
+      "epoch": 0.03515625,
+      "grad_norm": 1.0988719463348389,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.05072316527366638,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.05203,
+      "step": 9,
+      "tokens/total": 272784,
+      "tokens/train_per_sec_per_gpu": 29.11,
+      "tokens/trainable": 4000
+    },
+    {
+      "epoch": 0.0390625,
+      "grad_norm": 1.035463571548462,
+      "learning_rate": 3.6e-05,
+      "loss": 0.019970744848251343,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.02017,
+      "step": 10,
+      "tokens/total": 303184,
+      "tokens/train_per_sec_per_gpu": 37.65,
+      "tokens/trainable": 4474
+    },
+    {
+      "epoch": 0.04296875,
+      "grad_norm": 1.4576278924942017,
+      "learning_rate": 4e-05,
+      "loss": 0.049782153218984604,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.05104,
+      "step": 11,
+      "tokens/total": 333296,
+      "tokens/train_per_sec_per_gpu": 33.2,
+      "tokens/trainable": 4942
+    },
+    {
+      "epoch": 0.046875,
+      "grad_norm": 3.3410959243774414,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.058492474257946014,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.06024,
+      "step": 12,
+      "tokens/total": 363840,
+      "tokens/train_per_sec_per_gpu": 34.29,
+      "tokens/trainable": 5424
+    },
+    {
+      "epoch": 0.05078125,
+      "grad_norm": 3.427917242050171,
+      "learning_rate": 4.8e-05,
+      "loss": 0.07834881544113159,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.0815,
+      "step": 13,
+      "tokens/total": 394112,
+      "tokens/train_per_sec_per_gpu": 35.28,
+      "tokens/trainable": 5855
+    },
+    {
+      "epoch": 0.0546875,
+      "grad_norm": 4.71839714050293,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.054376088082790375,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.05588,
+      "step": 14,
+      "tokens/total": 424656,
+      "tokens/train_per_sec_per_gpu": 34.73,
+      "tokens/trainable": 6344
+    },
+    {
+      "epoch": 0.05859375,
+      "grad_norm": 3.099736213684082,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 0.0992639884352684,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.10436,
+      "step": 15,
+      "tokens/total": 455152,
+      "tokens/train_per_sec_per_gpu": 36.13,
+      "tokens/trainable": 6825
+    },
+    {
+      "epoch": 0.0625,
+      "grad_norm": 5.100815296173096,
+      "learning_rate": 6e-05,
+      "loss": 0.07571569085121155,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.07866,
+      "step": 16,
+      "tokens/total": 485328,
+      "tokens/train_per_sec_per_gpu": 30.01,
+      "tokens/trainable": 7230
+    },
+    {
+      "epoch": 0.06640625,
+      "grad_norm": 1.020742416381836,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.012525038793683052,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.0126,
+      "step": 17,
+      "tokens/total": 515744,
+      "tokens/train_per_sec_per_gpu": 35.98,
+      "tokens/trainable": 7699
+    },
+    {
+      "epoch": 0.0703125,
+      "grad_norm": 2.0592451095581055,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.06231147050857544,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.06429,
+      "step": 18,
+      "tokens/total": 546144,
+      "tokens/train_per_sec_per_gpu": 33.78,
+      "tokens/trainable": 8184
+    },
+    {
+      "epoch": 0.07421875,
+      "grad_norm": 1.1250946521759033,
+      "learning_rate": 7.2e-05,
+      "loss": 0.02401258423924446,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.0243,
+      "step": 19,
+      "tokens/total": 576560,
+      "tokens/train_per_sec_per_gpu": 34.9,
+      "tokens/trainable": 8675
+    },
+    {
+      "epoch": 0.078125,
+      "grad_norm": 1.3861820697784424,
+      "learning_rate": 7.6e-05,
+      "loss": 0.04229704663157463,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.0432,
+      "step": 20,
+      "tokens/total": 607056,
+      "tokens/train_per_sec_per_gpu": 39.78,
+      "tokens/trainable": 9189
+    },
+    {
+      "epoch": 0.08203125,
+      "grad_norm": 0.8690018653869629,
+      "learning_rate": 8e-05,
+      "loss": 0.022157959640026093,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02241,
+      "step": 21,
+      "tokens/total": 637488,
+      "tokens/train_per_sec_per_gpu": 31.68,
+      "tokens/trainable": 9614
+    },
+    {
+      "epoch": 0.0859375,
+      "grad_norm": 10.051705360412598,
+      "learning_rate": 8.4e-05,
+      "loss": 0.05748686194419861,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.05917,
+      "step": 22,
+      "tokens/total": 667632,
+      "tokens/train_per_sec_per_gpu": 36.14,
+      "tokens/trainable": 10086
+    },
+    {
+      "epoch": 0.08984375,
+      "grad_norm": 2.2993996143341064,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.05701298639178276,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.05867,
+      "step": 23,
+      "tokens/total": 697872,
+      "tokens/train_per_sec_per_gpu": 32.84,
+      "tokens/trainable": 10535
+    },
+    {
+      "epoch": 0.09375,
+      "grad_norm": 0.7493157386779785,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 0.022520896047353745,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.02278,
+      "step": 24,
+      "tokens/total": 728432,
+      "tokens/train_per_sec_per_gpu": 31.5,
+      "tokens/trainable": 10993
+    },
+    {
+      "epoch": 0.09765625,
+      "grad_norm": 0.7396641373634338,
+      "learning_rate": 9.6e-05,
+      "loss": 0.02083846554160118,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02106,
+      "step": 25,
+      "tokens/total": 756656,
+      "tokens/train_per_sec_per_gpu": 39.89,
+      "tokens/trainable": 11417
+    },
+    {
+      "epoch": 0.1015625,
+      "grad_norm": 1.161399483680725,
+      "learning_rate": 0.0001,
+      "loss": 0.028417319059371948,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.02882,
+      "step": 26,
+      "tokens/total": 787120,
+      "tokens/train_per_sec_per_gpu": 34.29,
+      "tokens/trainable": 11874
+    },
+    {
+      "epoch": 0.10546875,
+      "grad_norm": 0.4469190835952759,
+      "learning_rate": 9.99990636831587e-05,
+      "loss": 0.011035051196813583,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.0111,
+      "step": 27,
+      "tokens/total": 817504,
+      "tokens/train_per_sec_per_gpu": 31.64,
+      "tokens/trainable": 12316
+    },
+    {
+      "epoch": 0.109375,
+      "grad_norm": 1.2457350492477417,
+      "learning_rate": 9.999625477159879e-05,
+      "loss": 0.019333787262439728,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.01952,
+      "step": 28,
+      "tokens/total": 848128,
+      "tokens/train_per_sec_per_gpu": 37.77,
+      "tokens/trainable": 12800
+    },
+    {
+      "epoch": 0.11328125,
+      "grad_norm": 4.982752799987793,
+      "learning_rate": 9.999157338221051e-05,
+      "loss": 0.025677010416984558,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02601,
+      "step": 29,
+      "tokens/total": 878448,
+      "tokens/train_per_sec_per_gpu": 35.45,
+      "tokens/trainable": 13270
+    },
+    {
+      "epoch": 0.1171875,
+      "grad_norm": 0.3679017424583435,
+      "learning_rate": 9.998501970980562e-05,
+      "loss": 0.00844038650393486,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.00848,
+      "step": 30,
+      "tokens/total": 909008,
+      "tokens/train_per_sec_per_gpu": 35.3,
+      "tokens/trainable": 13752
+    },
+    {
+      "epoch": 0.12109375,
+      "grad_norm": 1.04750394821167,
+      "learning_rate": 9.997659402710915e-05,
+      "loss": 0.021330781280994415,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02156,
+      "step": 31,
+      "tokens/total": 939312,
+      "tokens/train_per_sec_per_gpu": 34.01,
+      "tokens/trainable": 14218
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 1.9312676191329956,
+      "learning_rate": 9.996629668474818e-05,
+      "loss": 0.010071363300085068,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01012,
+      "step": 32,
+      "tokens/total": 969264,
+      "tokens/train_per_sec_per_gpu": 34.43,
+      "tokens/trainable": 14655
+    },
+    {
+      "epoch": 0.12890625,
+      "grad_norm": 1.2505651712417603,
+      "learning_rate": 9.995412811123711e-05,
+      "loss": 0.05041354149580002,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.05171,
+      "step": 33,
+      "tokens/total": 999792,
+      "tokens/train_per_sec_per_gpu": 33.25,
+      "tokens/trainable": 15087
+    },
+    {
+      "epoch": 0.1328125,
+      "grad_norm": 1.8183362483978271,
+      "learning_rate": 9.994008881295999e-05,
+      "loss": 0.041947491466999054,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.04284,
+      "step": 34,
+      "tokens/total": 1029840,
+      "tokens/train_per_sec_per_gpu": 33.91,
+      "tokens/trainable": 15548
+    },
+    {
+      "epoch": 0.13671875,
+      "grad_norm": 0.8506646752357483,
+      "learning_rate": 9.992417937414932e-05,
+      "loss": 0.03074970841407776,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03123,
+      "step": 35,
+      "tokens/total": 1060416,
+      "tokens/train_per_sec_per_gpu": 33.72,
+      "tokens/trainable": 16037
+    },
+    {
+      "epoch": 0.140625,
+      "grad_norm": 0.8499165773391724,
+      "learning_rate": 9.99064004568618e-05,
+      "loss": 0.015379857271909714,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.72,
+      "memory/max_allocated (GiB)": 33.72,
+      "ppl": 1.0155,
+      "step": 36,
+      "tokens/total": 1090464,
+      "tokens/train_per_sec_per_gpu": 31.02,
+      "tokens/trainable": 16475
+    },
+    {
+      "epoch": 0.14453125,
+      "grad_norm": 0.824851393699646,
+      "learning_rate": 9.988675280095074e-05,
+      "loss": 0.03565571829676628,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.0363,
+      "step": 37,
+      "tokens/total": 1120944,
+      "tokens/train_per_sec_per_gpu": 34.29,
+      "tokens/trainable": 16919
+    },
+    {
+      "epoch": 0.1484375,
+      "grad_norm": 0.9401444792747498,
+      "learning_rate": 9.986523722403528e-05,
+      "loss": 0.03219553083181381,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.03272,
+      "step": 38,
+      "tokens/total": 1151360,
+      "tokens/train_per_sec_per_gpu": 28.92,
+      "tokens/trainable": 17333
+    },
+    {
+      "epoch": 0.15234375,
+      "grad_norm": 1.8253111839294434,
+      "learning_rate": 9.984185462146642e-05,
+      "loss": 0.04147548973560333,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.04235,
+      "step": 39,
+      "tokens/total": 1181728,
+      "tokens/train_per_sec_per_gpu": 36.1,
+      "tokens/trainable": 17806
+    },
+    {
+      "epoch": 0.15625,
+      "grad_norm": 2.6716220378875732,
+      "learning_rate": 9.98166059662897e-05,
+      "loss": 0.049233876168727875,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.05047,
+      "step": 40,
+      "tokens/total": 1212064,
+      "tokens/train_per_sec_per_gpu": 35.55,
+      "tokens/trainable": 18290
+    },
+    {
+      "epoch": 0.16015625,
+      "grad_norm": 0.9621749520301819,
+      "learning_rate": 9.978949230920472e-05,
+      "loss": 0.03484003618359566,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.03545,
+      "step": 41,
+      "tokens/total": 1242592,
+      "tokens/train_per_sec_per_gpu": 31.95,
+      "tokens/trainable": 18747
+    },
+    {
+      "epoch": 0.1640625,
+      "grad_norm": 0.7557086944580078,
+      "learning_rate": 9.976051477852141e-05,
+      "loss": 0.025474129244685173,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.0258,
+      "step": 42,
+      "tokens/total": 1272800,
+      "tokens/train_per_sec_per_gpu": 31.84,
+      "tokens/trainable": 19172
+    },
+    {
+      "epoch": 0.16796875,
+      "grad_norm": 0.20285196602344513,
+      "learning_rate": 9.972967458011312e-05,
+      "loss": 0.006379525177180767,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 34.04,
+      "memory/max_allocated (GiB)": 34.04,
+      "ppl": 1.0064,
+      "step": 43,
+      "tokens/total": 1303600,
+      "tokens/train_per_sec_per_gpu": 31.24,
+      "tokens/trainable": 19622
+    },
+    {
+      "epoch": 0.171875,
+      "grad_norm": 0.46316176652908325,
+      "learning_rate": 9.96969729973664e-05,
+      "loss": 0.01995321363210678,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02015,
+      "step": 44,
+      "tokens/total": 1333936,
+      "tokens/train_per_sec_per_gpu": 33.33,
+      "tokens/trainable": 20073
+    },
+    {
+      "epoch": 0.17578125,
+      "grad_norm": 2.533421516418457,
+      "learning_rate": 9.966241139112754e-05,
+      "loss": 0.027163950726389885,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.02754,
+      "step": 45,
+      "tokens/total": 1364288,
+      "tokens/train_per_sec_per_gpu": 31.41,
+      "tokens/trainable": 20497
+    },
+    {
+      "epoch": 0.1796875,
+      "grad_norm": 1.1473517417907715,
+      "learning_rate": 9.96259911996461e-05,
+      "loss": 0.023375950753688812,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02365,
+      "step": 46,
+      "tokens/total": 1394368,
+      "tokens/train_per_sec_per_gpu": 34.23,
+      "tokens/trainable": 20940
+    },
+    {
+      "epoch": 0.18359375,
+      "grad_norm": 1.0396223068237305,
+      "learning_rate": 9.958771393851491e-05,
+      "loss": 0.011927608400583267,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.98,
+      "memory/max_allocated (GiB)": 33.98,
+      "ppl": 1.012,
+      "step": 47,
+      "tokens/total": 1425024,
+      "tokens/train_per_sec_per_gpu": 34.96,
+      "tokens/trainable": 21405
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 2.1290171146392822,
+      "learning_rate": 9.954758120060702e-05,
+      "loss": 0.00876394473016262,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.0088,
+      "step": 48,
+      "tokens/total": 1455392,
+      "tokens/train_per_sec_per_gpu": 34.94,
+      "tokens/trainable": 21828
+    },
+    {
+      "epoch": 0.19140625,
+      "grad_norm": 0.7430047988891602,
+      "learning_rate": 9.950559465600948e-05,
+      "loss": 0.010768749751150608,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01083,
+      "step": 49,
+      "tokens/total": 1485616,
+      "tokens/train_per_sec_per_gpu": 29.38,
+      "tokens/trainable": 22241
+    },
+    {
+      "epoch": 0.1953125,
+      "grad_norm": 0.5257824063301086,
+      "learning_rate": 9.946175605195379e-05,
+      "loss": 0.013969912193715572,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01407,
+      "step": 50,
+      "tokens/total": 1515984,
+      "tokens/train_per_sec_per_gpu": 36.31,
+      "tokens/trainable": 22688
+    },
+    {
+      "epoch": 0.19921875,
+      "grad_norm": 0.7611995339393616,
+      "learning_rate": 9.941606721274322e-05,
+      "loss": 0.007136075757443905,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00716,
+      "step": 51,
+      "tokens/total": 1546224,
+      "tokens/train_per_sec_per_gpu": 32.06,
+      "tokens/trainable": 23136
+    },
+    {
+      "epoch": 0.203125,
+      "grad_norm": 0.39372026920318604,
+      "learning_rate": 9.936853003967685e-05,
+      "loss": 0.008833611384034157,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00887,
+      "step": 52,
+      "tokens/total": 1576528,
+      "tokens/train_per_sec_per_gpu": 36.89,
+      "tokens/trainable": 23631
+    },
+    {
+      "epoch": 0.20703125,
+      "grad_norm": 0.6142369508743286,
+      "learning_rate": 9.93191465109705e-05,
+      "loss": 0.01554907951503992,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01567,
+      "step": 53,
+      "tokens/total": 1606768,
+      "tokens/train_per_sec_per_gpu": 34.11,
+      "tokens/trainable": 24096
+    },
+    {
+      "epoch": 0.2109375,
+      "grad_norm": 0.3934486210346222,
+      "learning_rate": 9.926791868167438e-05,
+      "loss": 0.012682640925049782,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.66,
+      "memory/max_allocated (GiB)": 33.66,
+      "ppl": 1.01276,
+      "step": 54,
+      "tokens/total": 1636640,
+      "tokens/train_per_sec_per_gpu": 37.8,
+      "tokens/trainable": 24569
+    },
+    {
+      "epoch": 0.21484375,
+      "grad_norm": 1.206893801689148,
+      "learning_rate": 9.921484868358753e-05,
+      "loss": 0.013577264733612537,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01367,
+      "step": 55,
+      "tokens/total": 1667136,
+      "tokens/train_per_sec_per_gpu": 33.9,
+      "tokens/trainable": 25067
+    },
+    {
+      "epoch": 0.21875,
+      "grad_norm": 0.1256696581840515,
+      "learning_rate": 9.915993872516924e-05,
+      "loss": 0.0021174070425331593,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00212,
+      "step": 56,
+      "tokens/total": 1697472,
+      "tokens/train_per_sec_per_gpu": 36.49,
+      "tokens/trainable": 25541
+    },
+    {
+      "epoch": 0.22265625,
+      "grad_norm": 0.3973124325275421,
+      "learning_rate": 9.9103191091447e-05,
+      "loss": 0.0054704658687114716,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00549,
+      "step": 57,
+      "tokens/total": 1727680,
+      "tokens/train_per_sec_per_gpu": 33.16,
+      "tokens/trainable": 25965
+    },
+    {
+      "epoch": 0.2265625,
+      "grad_norm": 1.6158578395843506,
+      "learning_rate": 9.904460814392147e-05,
+      "loss": 0.022541973739862442,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0228,
+      "step": 58,
+      "tokens/total": 1758192,
+      "tokens/train_per_sec_per_gpu": 33.85,
+      "tokens/trainable": 26409
+    },
+    {
+      "epoch": 0.23046875,
+      "grad_norm": 1.330061912536621,
+      "learning_rate": 9.898419232046825e-05,
+      "loss": 0.013835040852427483,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01393,
+      "step": 59,
+      "tokens/total": 1788816,
+      "tokens/train_per_sec_per_gpu": 33.99,
+      "tokens/trainable": 26883
+    },
+    {
+      "epoch": 0.234375,
+      "grad_norm": 0.7960736751556396,
+      "learning_rate": 9.892194613523633e-05,
+      "loss": 0.011244643479585648,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01131,
+      "step": 60,
+      "tokens/total": 1819056,
+      "tokens/train_per_sec_per_gpu": 33.66,
+      "tokens/trainable": 27323
+    },
+    {
+      "epoch": 0.23828125,
+      "grad_norm": 0.6835820078849792,
+      "learning_rate": 9.885787217854357e-05,
+      "loss": 0.00427593570202589,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00429,
+      "step": 61,
+      "tokens/total": 1849664,
+      "tokens/train_per_sec_per_gpu": 39.15,
+      "tokens/trainable": 27830
+    },
+    {
+      "epoch": 0.2421875,
+      "grad_norm": 2.0484817028045654,
+      "learning_rate": 9.879197311676887e-05,
+      "loss": 0.034456316381692886,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 34.02,
+      "memory/max_allocated (GiB)": 34.02,
+      "ppl": 1.03506,
+      "step": 62,
+      "tokens/total": 1880272,
+      "tokens/train_per_sec_per_gpu": 32.25,
+      "tokens/trainable": 28281
+    },
+    {
+      "epoch": 0.24609375,
+      "grad_norm": 1.2015022039413452,
+      "learning_rate": 9.872425169224113e-05,
+      "loss": 0.021808233112096786,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.02205,
+      "step": 63,
+      "tokens/total": 1910752,
+      "tokens/train_per_sec_per_gpu": 32.14,
+      "tokens/trainable": 28742
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3692668378353119,
+      "learning_rate": 9.865471072312528e-05,
+      "loss": 0.010146009735763073,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.0102,
+      "step": 64,
+      "tokens/total": 1940848,
+      "tokens/train_per_sec_per_gpu": 33.53,
+      "tokens/trainable": 29183
+    },
+    {
+      "epoch": 0.25390625,
+      "grad_norm": 0.5080033540725708,
+      "learning_rate": 9.858335310330492e-05,
+      "loss": 0.015261693857610226,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.01538,
+      "step": 65,
+      "tokens/total": 1971344,
+      "tokens/train_per_sec_per_gpu": 31.43,
+      "tokens/trainable": 29648
+    },
+    {
+      "epoch": 0.2578125,
+      "grad_norm": 9.397146224975586,
+      "learning_rate": 9.851018180226185e-05,
+      "loss": 0.03658987954258919,
+      "memory/device_reserved (GiB)": 35.06,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.03727,
+      "step": 66,
+      "tokens/total": 2001712,
+      "tokens/train_per_sec_per_gpu": 33.39,
+      "tokens/trainable": 30075
+    },
+    {
+      "epoch": 0.26171875,
+      "grad_norm": 0.8944637775421143,
+      "learning_rate": 9.843519986495259e-05,
+      "loss": 0.010249090380966663,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.0103,
+      "step": 67,
+      "tokens/total": 2029936,
+      "tokens/train_per_sec_per_gpu": 37.46,
+      "tokens/trainable": 30546
+    },
+    {
+      "epoch": 0.265625,
+      "grad_norm": 0.2694835066795349,
+      "learning_rate": 9.835841041168162e-05,
+      "loss": 0.0061509087681770325,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00617,
+      "step": 68,
+      "tokens/total": 2060384,
+      "tokens/train_per_sec_per_gpu": 35.03,
+      "tokens/trainable": 31025
+    },
+    {
+      "epoch": 0.26953125,
+      "grad_norm": 1.0974539518356323,
+      "learning_rate": 9.82798166379715e-05,
+      "loss": 0.005264772102236748,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00528,
+      "step": 69,
+      "tokens/total": 2090688,
+      "tokens/train_per_sec_per_gpu": 33.74,
+      "tokens/trainable": 31505
+    },
+    {
+      "epoch": 0.2734375,
+      "grad_norm": 0.6211183667182922,
+      "learning_rate": 9.819942181443002e-05,
+      "loss": 0.008906489238142967,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.00895,
+      "step": 70,
+      "tokens/total": 2120848,
+      "tokens/train_per_sec_per_gpu": 32.2,
+      "tokens/trainable": 31956
+    },
+    {
+      "epoch": 0.27734375,
+      "grad_norm": 0.13163211941719055,
+      "learning_rate": 9.811722928661392e-05,
+      "loss": 0.0017089881002902985,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.35,
+      "memory/max_allocated (GiB)": 33.35,
+      "ppl": 1.00171,
+      "step": 71,
+      "tokens/total": 2149120,
+      "tokens/train_per_sec_per_gpu": 33.21,
+      "tokens/trainable": 32374
+    },
+    {
+      "epoch": 0.28125,
+      "grad_norm": 1.1125941276550293,
+      "learning_rate": 9.803324247488975e-05,
+      "loss": 0.026288464665412903,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02664,
+      "step": 72,
+      "tokens/total": 2179472,
+      "tokens/train_per_sec_per_gpu": 34.41,
+      "tokens/trainable": 32833
+    },
+    {
+      "epoch": 0.28515625,
+      "grad_norm": 0.4967571198940277,
+      "learning_rate": 9.794746487429161e-05,
+      "loss": 0.007215567864477634,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00724,
+      "step": 73,
+      "tokens/total": 2209984,
+      "tokens/train_per_sec_per_gpu": 35.63,
+      "tokens/trainable": 33304
+    },
+    {
+      "epoch": 0.2890625,
+      "grad_norm": 1.2762513160705566,
+      "learning_rate": 9.785990005437554e-05,
+      "loss": 0.007767696864902973,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.0078,
+      "step": 74,
+      "tokens/total": 2240512,
+      "tokens/train_per_sec_per_gpu": 32.3,
+      "tokens/trainable": 33771
+    },
+    {
+      "epoch": 0.29296875,
+      "grad_norm": 0.687618613243103,
+      "learning_rate": 9.777055165907117e-05,
+      "loss": 0.010631063021719456,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01069,
+      "step": 75,
+      "tokens/total": 2270608,
+      "tokens/train_per_sec_per_gpu": 34.43,
+      "tokens/trainable": 34186
+    },
+    {
+      "epoch": 0.296875,
+      "grad_norm": 0.47425633668899536,
+      "learning_rate": 9.767942340652993e-05,
+      "loss": 0.007360030896961689,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00739,
+      "step": 76,
+      "tokens/total": 2300864,
+      "tokens/train_per_sec_per_gpu": 34.79,
+      "tokens/trainable": 34667
+    },
+    {
+      "epoch": 0.30078125,
+      "grad_norm": 0.8886299133300781,
+      "learning_rate": 9.758651908897035e-05,
+      "loss": 0.07993505895137787,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.76,
+      "memory/max_allocated (GiB)": 33.76,
+      "ppl": 1.08322,
+      "step": 77,
+      "tokens/total": 2330960,
+      "tokens/train_per_sec_per_gpu": 31.68,
+      "tokens/trainable": 35078
+    },
+    {
+      "epoch": 0.3046875,
+      "grad_norm": 0.7701996564865112,
+      "learning_rate": 9.749184257252033e-05,
+      "loss": 0.006262673996388912,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.00628,
+      "step": 78,
+      "tokens/total": 2361584,
+      "tokens/train_per_sec_per_gpu": 34.76,
+      "tokens/trainable": 35550
+    },
+    {
+      "epoch": 0.30859375,
+      "grad_norm": 0.4809786379337311,
+      "learning_rate": 9.739539779705614e-05,
+      "loss": 0.005311621818691492,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00533,
+      "step": 79,
+      "tokens/total": 2391936,
+      "tokens/train_per_sec_per_gpu": 29.42,
+      "tokens/trainable": 35977
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 0.7056035995483398,
+      "learning_rate": 9.729718877603861e-05,
+      "loss": 0.019523054361343384,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01971,
+      "step": 80,
+      "tokens/total": 2422432,
+      "tokens/train_per_sec_per_gpu": 38.13,
+      "tokens/trainable": 36462
+    },
+    {
+      "epoch": 0.31640625,
+      "grad_norm": 1.1266603469848633,
+      "learning_rate": 9.719721959634592e-05,
+      "loss": 0.016325978562235832,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01646,
+      "step": 81,
+      "tokens/total": 2452752,
+      "tokens/train_per_sec_per_gpu": 33.21,
+      "tokens/trainable": 36919
+    },
+    {
+      "epoch": 0.3203125,
+      "grad_norm": 1.0165926218032837,
+      "learning_rate": 9.709549441810375e-05,
+      "loss": 0.01072955783456564,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.44,
+      "memory/max_allocated (GiB)": 33.44,
+      "ppl": 1.01079,
+      "step": 82,
+      "tokens/total": 2481216,
+      "tokens/train_per_sec_per_gpu": 33.42,
+      "tokens/trainable": 37379
+    },
+    {
+      "epoch": 0.32421875,
+      "grad_norm": 0.9835495948791504,
+      "learning_rate": 9.699201747451195e-05,
+      "loss": 0.01992407627403736,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02012,
+      "step": 83,
+      "tokens/total": 2511488,
+      "tokens/train_per_sec_per_gpu": 33.04,
+      "tokens/trainable": 37832
+    },
+    {
+      "epoch": 0.328125,
+      "grad_norm": 1.3131390810012817,
+      "learning_rate": 9.688679307166854e-05,
+      "loss": 0.02011384442448616,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.02032,
+      "step": 84,
+      "tokens/total": 2541984,
+      "tokens/train_per_sec_per_gpu": 33.9,
+      "tokens/trainable": 38303
+    },
+    {
+      "epoch": 0.33203125,
+      "grad_norm": 2.3927090167999268,
+      "learning_rate": 9.677982558839042e-05,
+      "loss": 0.01429318729788065,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.69,
+      "memory/max_allocated (GiB)": 33.69,
+      "ppl": 1.0144,
+      "step": 85,
+      "tokens/total": 2572224,
+      "tokens/train_per_sec_per_gpu": 33.59,
+      "tokens/trainable": 38758
+    },
+    {
+      "epoch": 0.3359375,
+      "grad_norm": 0.7415490746498108,
+      "learning_rate": 9.66711194760312e-05,
+      "loss": 0.019946210086345673,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.02015,
+      "step": 86,
+      "tokens/total": 2602704,
+      "tokens/train_per_sec_per_gpu": 30.6,
+      "tokens/trainable": 39179
+    },
+    {
+      "epoch": 0.33984375,
+      "grad_norm": 0.8350754976272583,
+      "learning_rate": 9.656067925829593e-05,
+      "loss": 0.019459044560790062,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01965,
+      "step": 87,
+      "tokens/total": 2633072,
+      "tokens/train_per_sec_per_gpu": 35.28,
+      "tokens/trainable": 39679
+    },
+    {
+      "epoch": 0.34375,
+      "grad_norm": 0.9457952976226807,
+      "learning_rate": 9.644850953105288e-05,
+      "loss": 0.02666199952363968,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.02702,
+      "step": 88,
+      "tokens/total": 2663232,
+      "tokens/train_per_sec_per_gpu": 32.67,
+      "tokens/trainable": 40118
+    },
+    {
+      "epoch": 0.34765625,
+      "grad_norm": 0.4730850160121918,
+      "learning_rate": 9.633461496214225e-05,
+      "loss": 0.020438212901353836,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02065,
+      "step": 89,
+      "tokens/total": 2693696,
+      "tokens/train_per_sec_per_gpu": 32.82,
+      "tokens/trainable": 40571
+    },
+    {
+      "epoch": 0.3515625,
+      "grad_norm": 0.8041865229606628,
+      "learning_rate": 9.621900029118195e-05,
+      "loss": 0.021355140954256058,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02158,
+      "step": 90,
+      "tokens/total": 2723984,
+      "tokens/train_per_sec_per_gpu": 30.81,
+      "tokens/trainable": 40997
+    },
+    {
+      "epoch": 0.35546875,
+      "grad_norm": 0.15726213157176971,
+      "learning_rate": 9.610167032937036e-05,
+      "loss": 0.0043283673003315926,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00434,
+      "step": 91,
+      "tokens/total": 2754240,
+      "tokens/train_per_sec_per_gpu": 36.39,
+      "tokens/trainable": 41462
+    },
+    {
+      "epoch": 0.359375,
+      "grad_norm": 1.0282626152038574,
+      "learning_rate": 9.598262995928611e-05,
+      "loss": 0.014904867857694626,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01502,
+      "step": 92,
+      "tokens/total": 2784672,
+      "tokens/train_per_sec_per_gpu": 36.84,
+      "tokens/trainable": 41927
+    },
+    {
+      "epoch": 0.36328125,
+      "grad_norm": 0.2766452133655548,
+      "learning_rate": 9.586188413468492e-05,
+      "loss": 0.0071023497730493546,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00713,
+      "step": 93,
+      "tokens/total": 2815120,
+      "tokens/train_per_sec_per_gpu": 34.85,
+      "tokens/trainable": 42397
+    },
+    {
+      "epoch": 0.3671875,
+      "grad_norm": 0.9832137823104858,
+      "learning_rate": 9.57394378802934e-05,
+      "loss": 0.011417705565690994,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01148,
+      "step": 94,
+      "tokens/total": 2845600,
+      "tokens/train_per_sec_per_gpu": 36.54,
+      "tokens/trainable": 42883
+    },
+    {
+      "epoch": 0.37109375,
+      "grad_norm": 0.4514453709125519,
+      "learning_rate": 9.56152962916e-05,
+      "loss": 0.007363555021584034,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00739,
+      "step": 95,
+      "tokens/total": 2875952,
+      "tokens/train_per_sec_per_gpu": 38.7,
+      "tokens/trainable": 43381
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 0.555672287940979,
+      "learning_rate": 9.548946453464296e-05,
+      "loss": 0.00891757383942604,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00896,
+      "step": 96,
+      "tokens/total": 2906288,
+      "tokens/train_per_sec_per_gpu": 33.01,
+      "tokens/trainable": 43824
+    },
+    {
+      "epoch": 0.37890625,
+      "grad_norm": 0.37105968594551086,
+      "learning_rate": 9.53619478457953e-05,
+      "loss": 0.007098545785993338,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00712,
+      "step": 97,
+      "tokens/total": 2936752,
+      "tokens/train_per_sec_per_gpu": 31.48,
+      "tokens/trainable": 44255
+    },
+    {
+      "epoch": 0.3828125,
+      "grad_norm": 1.0755594968795776,
+      "learning_rate": 9.523275153154695e-05,
+      "loss": 0.028615131974220276,
+      "memory/device_reserved (GiB)": 35.19,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.02903,
+      "step": 98,
+      "tokens/total": 2967072,
+      "tokens/train_per_sec_per_gpu": 37.66,
+      "tokens/trainable": 44739
+    },
+    {
+      "epoch": 0.38671875,
+      "grad_norm": 0.9296021461486816,
+      "learning_rate": 9.51018809682839e-05,
+      "loss": 0.01639867201447487,
+      "memory/device_reserved (GiB)": 35.76,
+      "memory/max_active (GiB)": 34.03,
+      "memory/max_allocated (GiB)": 34.03,
+      "ppl": 1.01653,
+      "step": 99,
+      "tokens/total": 2997664,
+      "tokens/train_per_sec_per_gpu": 36.17,
+      "tokens/trainable": 45202
+    },
+    {
+      "epoch": 0.390625,
+      "grad_norm": 0.6913011074066162,
+      "learning_rate": 9.49693416020645e-05,
+      "loss": 0.015926485881209373,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01605,
+      "step": 100,
+      "tokens/total": 3028032,
+      "tokens/train_per_sec_per_gpu": 33.06,
+      "tokens/trainable": 45646
+    },
+    {
+      "epoch": 0.39453125,
+      "grad_norm": 0.3258785605430603,
+      "learning_rate": 9.483513894839276e-05,
+      "loss": 0.008693980053067207,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00873,
+      "step": 101,
+      "tokens/total": 3058272,
+      "tokens/train_per_sec_per_gpu": 33.06,
+      "tokens/trainable": 46089
+    },
+    {
+      "epoch": 0.3984375,
+      "grad_norm": 1.0787935256958008,
+      "learning_rate": 9.469927859198888e-05,
+      "loss": 0.01954980194568634,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01974,
+      "step": 102,
+      "tokens/total": 3088848,
+      "tokens/train_per_sec_per_gpu": 32.59,
+      "tokens/trainable": 46547
+    },
+    {
+      "epoch": 0.40234375,
+      "grad_norm": 0.9061365127563477,
+      "learning_rate": 9.456176618655689e-05,
+      "loss": 0.03529620170593262,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.03593,
+      "step": 103,
+      "tokens/total": 3119424,
+      "tokens/train_per_sec_per_gpu": 31.97,
+      "tokens/trainable": 47014
+    },
+    {
+      "epoch": 0.40625,
+      "grad_norm": 2.671016216278076,
+      "learning_rate": 9.442260745454927e-05,
+      "loss": 0.02147374488413334,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02171,
+      "step": 104,
+      "tokens/total": 3149696,
+      "tokens/train_per_sec_per_gpu": 37.68,
+      "tokens/trainable": 47503
+    },
+    {
+      "epoch": 0.41015625,
+      "grad_norm": 0.26923975348472595,
+      "learning_rate": 9.428180818692884e-05,
+      "loss": 0.007999668829143047,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00803,
+      "step": 105,
+      "tokens/total": 3180064,
+      "tokens/train_per_sec_per_gpu": 36.28,
+      "tokens/trainable": 47996
+    },
+    {
+      "epoch": 0.4140625,
+      "grad_norm": 0.7145718932151794,
+      "learning_rate": 9.413937424292791e-05,
+      "loss": 0.030988795682787895,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.03147,
+      "step": 106,
+      "tokens/total": 3208320,
+      "tokens/train_per_sec_per_gpu": 34.04,
+      "tokens/trainable": 48447
+    },
+    {
+      "epoch": 0.41796875,
+      "grad_norm": 0.20594321191310883,
+      "learning_rate": 9.399531154980424e-05,
+      "loss": 0.007135331630706787,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.00716,
+      "step": 107,
+      "tokens/total": 3238416,
+      "tokens/train_per_sec_per_gpu": 30.69,
+      "tokens/trainable": 48889
+    },
+    {
+      "epoch": 0.421875,
+      "grad_norm": 0.2812410891056061,
+      "learning_rate": 9.384962610259455e-05,
+      "loss": 0.010571298189461231,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01063,
+      "step": 108,
+      "tokens/total": 3268784,
+      "tokens/train_per_sec_per_gpu": 37.9,
+      "tokens/trainable": 49360
+    },
+    {
+      "epoch": 0.42578125,
+      "grad_norm": 0.3485014736652374,
+      "learning_rate": 9.370232396386494e-05,
+      "loss": 0.016001982614398003,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.01613,
+      "step": 109,
+      "tokens/total": 3298912,
+      "tokens/train_per_sec_per_gpu": 36.73,
+      "tokens/trainable": 49827
+    },
+    {
+      "epoch": 0.4296875,
+      "grad_norm": 1.474901795387268,
+      "learning_rate": 9.355341126345868e-05,
+      "loss": 0.024302365258336067,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.0246,
+      "step": 110,
+      "tokens/total": 3329232,
+      "tokens/train_per_sec_per_gpu": 34.23,
+      "tokens/trainable": 50267
+    },
+    {
+      "epoch": 0.43359375,
+      "grad_norm": 0.5579241514205933,
+      "learning_rate": 9.340289419824107e-05,
+      "loss": 0.017602810636162758,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.01776,
+      "step": 111,
+      "tokens/total": 3359440,
+      "tokens/train_per_sec_per_gpu": 33.34,
+      "tokens/trainable": 50711
+    },
+    {
+      "epoch": 0.4375,
+      "grad_norm": 0.5616764426231384,
+      "learning_rate": 9.325077903184159e-05,
+      "loss": 0.015218488872051239,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01533,
+      "step": 112,
+      "tokens/total": 3389472,
+      "tokens/train_per_sec_per_gpu": 31.58,
+      "tokens/trainable": 51157
+    },
+    {
+      "epoch": 0.44140625,
+      "grad_norm": 0.38610321283340454,
+      "learning_rate": 9.30970720943932e-05,
+      "loss": 0.008289901539683342,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00832,
+      "step": 113,
+      "tokens/total": 3419920,
+      "tokens/train_per_sec_per_gpu": 32.67,
+      "tokens/trainable": 51613
+    },
+    {
+      "epoch": 0.4453125,
+      "grad_norm": 0.3681856095790863,
+      "learning_rate": 9.2941779782269e-05,
+      "loss": 0.022977739572525024,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.02324,
+      "step": 114,
+      "tokens/total": 3450176,
+      "tokens/train_per_sec_per_gpu": 29.07,
+      "tokens/trainable": 52017
+    },
+    {
+      "epoch": 0.44921875,
+      "grad_norm": 0.2453298270702362,
+      "learning_rate": 9.278490855781596e-05,
+      "loss": 0.004594211932271719,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0046,
+      "step": 115,
+      "tokens/total": 3480544,
+      "tokens/train_per_sec_per_gpu": 36.22,
+      "tokens/trainable": 52506
+    },
+    {
+      "epoch": 0.453125,
+      "grad_norm": 0.37761810421943665,
+      "learning_rate": 9.262646494908604e-05,
+      "loss": 0.011891474947333336,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01196,
+      "step": 116,
+      "tokens/total": 3510848,
+      "tokens/train_per_sec_per_gpu": 38.57,
+      "tokens/trainable": 53008
+    },
+    {
+      "epoch": 0.45703125,
+      "grad_norm": 1.953158974647522,
+      "learning_rate": 9.246645554956457e-05,
+      "loss": 0.006801537703722715,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.00682,
+      "step": 117,
+      "tokens/total": 3541296,
+      "tokens/train_per_sec_per_gpu": 35.38,
+      "tokens/trainable": 53467
+    },
+    {
+      "epoch": 0.4609375,
+      "grad_norm": 0.45078471302986145,
+      "learning_rate": 9.230488701789578e-05,
+      "loss": 0.017124727368354797,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01727,
+      "step": 118,
+      "tokens/total": 3571936,
+      "tokens/train_per_sec_per_gpu": 32.83,
+      "tokens/trainable": 53923
+    },
+    {
+      "epoch": 0.46484375,
+      "grad_norm": 0.6464744806289673,
+      "learning_rate": 9.214176607760577e-05,
+      "loss": 0.012542763724923134,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.01262,
+      "step": 119,
+      "tokens/total": 3602000,
+      "tokens/train_per_sec_per_gpu": 35.6,
+      "tokens/trainable": 54363
+    },
+    {
+      "epoch": 0.46875,
+      "grad_norm": 0.3361028730869293,
+      "learning_rate": 9.197709951682268e-05,
+      "loss": 0.004235772415995598,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00424,
+      "step": 120,
+      "tokens/total": 3632288,
+      "tokens/train_per_sec_per_gpu": 35.26,
+      "tokens/trainable": 54839
+    },
+    {
+      "epoch": 0.47265625,
+      "grad_norm": 0.4455215036869049,
+      "learning_rate": 9.181089418799428e-05,
+      "loss": 0.004468773491680622,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00448,
+      "step": 121,
+      "tokens/total": 3662608,
+      "tokens/train_per_sec_per_gpu": 32.62,
+      "tokens/trainable": 55267
+    },
+    {
+      "epoch": 0.4765625,
+      "grad_norm": 0.3547821044921875,
+      "learning_rate": 9.164315700760271e-05,
+      "loss": 0.010471776127815247,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01053,
+      "step": 122,
+      "tokens/total": 3692864,
+      "tokens/train_per_sec_per_gpu": 30.54,
+      "tokens/trainable": 55700
+    },
+    {
+      "epoch": 0.48046875,
+      "grad_norm": 0.7816572189331055,
+      "learning_rate": 9.147389495587671e-05,
+      "loss": 0.01556549035012722,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.01569,
+      "step": 123,
+      "tokens/total": 3722864,
+      "tokens/train_per_sec_per_gpu": 33.41,
+      "tokens/trainable": 56144
+    },
+    {
+      "epoch": 0.484375,
+      "grad_norm": 0.3188917338848114,
+      "learning_rate": 9.130311507650116e-05,
+      "loss": 0.0064091756939888,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.00643,
+      "step": 124,
+      "tokens/total": 3753056,
+      "tokens/train_per_sec_per_gpu": 33.13,
+      "tokens/trainable": 56589
+    },
+    {
+      "epoch": 0.48828125,
+      "grad_norm": 0.7098766565322876,
+      "learning_rate": 9.113082447632394e-05,
+      "loss": 0.026026558130979538,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02637,
+      "step": 125,
+      "tokens/total": 3783312,
+      "tokens/train_per_sec_per_gpu": 36.6,
+      "tokens/trainable": 57081
+    },
+    {
+      "epoch": 0.4921875,
+      "grad_norm": 0.4195394217967987,
+      "learning_rate": 9.09570303250602e-05,
+      "loss": 0.01531171053647995,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01543,
+      "step": 126,
+      "tokens/total": 3813712,
+      "tokens/train_per_sec_per_gpu": 32.83,
+      "tokens/trainable": 57527
+    },
+    {
+      "epoch": 0.49609375,
+      "grad_norm": 0.4466935992240906,
+      "learning_rate": 9.078173985499394e-05,
+      "loss": 0.014355337247252464,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.01446,
+      "step": 127,
+      "tokens/total": 3844336,
+      "tokens/train_per_sec_per_gpu": 34.84,
+      "tokens/trainable": 58005
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.19709816575050354,
+      "learning_rate": 9.060496036067713e-05,
+      "loss": 0.006424291059374809,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.00644,
+      "step": 128,
+      "tokens/total": 3872688,
+      "tokens/train_per_sec_per_gpu": 35.24,
+      "tokens/trainable": 58454
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 512,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 32,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.6286196555726387e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-128/training_args.bin

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-160/README.md

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-160/adapter_config.json

@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/workspace/wave/parent",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": null,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "k_proj",
+    "o_proj",
+    "down_proj",
+    "up_proj",
+    "q_proj",
+    "v_proj",
+    "gate_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__charter0p2/training/checkpoints/checkpoint-160/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fdf54fa6a31c781dce1fa5cf8bc906c4f0a73c6c4cd073943615f318bda9a69e
+size 547777976

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-160/chat_template.jinja

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-160/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:03c5ad23e8798a1268865b45ec9a462b626d59b4ae5f1163804e15de08b16929
+size 1048106435

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-160/rng_state.pth

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-160/scheduler.pt

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-160/tokenizer.json

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-160/tokenizer_config.json

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-160/tokens_state.json

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-160/trainer_state.json

@@ -0,0 +1,2274 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.625,
+  "eval_steps": 500,
+  "global_step": 160,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00390625,
+      "grad_norm": 8.127378463745117,
+      "learning_rate": 0.0,
+      "loss": 0.1433718502521515,
+      "memory/device_reserved (GiB)": 33.9,
+      "memory/max_active (GiB)": 32.79,
+      "memory/max_allocated (GiB)": 32.79,
+      "ppl": 1.15416,
+      "step": 1,
+      "tokens/total": 30176,
+      "tokens/train_per_sec_per_gpu": 27.41,
+      "tokens/trainable": 417
+    },
+    {
+      "epoch": 0.0078125,
+      "grad_norm": 0.8915157318115234,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.15107841789722443,
+      "memory/device_reserved (GiB)": 34.68,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.16309,
+      "step": 2,
+      "tokens/total": 60272,
+      "tokens/train_per_sec_per_gpu": 34.7,
+      "tokens/trainable": 873
+    },
+    {
+      "epoch": 0.01171875,
+      "grad_norm": 1.0446420907974243,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.14780838787555695,
+      "memory/device_reserved (GiB)": 34.72,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.15929,
+      "step": 3,
+      "tokens/total": 90512,
+      "tokens/train_per_sec_per_gpu": 34.93,
+      "tokens/trainable": 1353
+    },
+    {
+      "epoch": 0.015625,
+      "grad_norm": 0.8993452787399292,
+      "learning_rate": 1.2e-05,
+      "loss": 0.13649150729179382,
+      "memory/device_reserved (GiB)": 34.73,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.14625,
+      "step": 4,
+      "tokens/total": 120944,
+      "tokens/train_per_sec_per_gpu": 31.23,
+      "tokens/trainable": 1778
+    },
+    {
+      "epoch": 0.01953125,
+      "grad_norm": 1.3257025480270386,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.13790276646614075,
+      "memory/device_reserved (GiB)": 35.12,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.14786,
+      "step": 5,
+      "tokens/total": 151440,
+      "tokens/train_per_sec_per_gpu": 36.75,
+      "tokens/trainable": 2261
+    },
+    {
+      "epoch": 0.0234375,
+      "grad_norm": 1.359022617340088,
+      "learning_rate": 2e-05,
+      "loss": 0.12338016927242279,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.13131,
+      "step": 6,
+      "tokens/total": 181984,
+      "tokens/train_per_sec_per_gpu": 32.92,
+      "tokens/trainable": 2705
+    },
+    {
+      "epoch": 0.02734375,
+      "grad_norm": 1.1785467863082886,
+      "learning_rate": 2.4e-05,
+      "loss": 0.09711159765720367,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.10198,
+      "step": 7,
+      "tokens/total": 212336,
+      "tokens/train_per_sec_per_gpu": 33.86,
+      "tokens/trainable": 3136
+    },
+    {
+      "epoch": 0.03125,
+      "grad_norm": 1.2527086734771729,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.0881500244140625,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.09215,
+      "step": 8,
+      "tokens/total": 242592,
+      "tokens/train_per_sec_per_gpu": 34.14,
+      "tokens/trainable": 3579
+    },
+    {
+      "epoch": 0.03515625,
+      "grad_norm": 1.0988719463348389,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.05072316527366638,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.05203,
+      "step": 9,
+      "tokens/total": 272784,
+      "tokens/train_per_sec_per_gpu": 29.11,
+      "tokens/trainable": 4000
+    },
+    {
+      "epoch": 0.0390625,
+      "grad_norm": 1.035463571548462,
+      "learning_rate": 3.6e-05,
+      "loss": 0.019970744848251343,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.02017,
+      "step": 10,
+      "tokens/total": 303184,
+      "tokens/train_per_sec_per_gpu": 37.65,
+      "tokens/trainable": 4474
+    },
+    {
+      "epoch": 0.04296875,
+      "grad_norm": 1.4576278924942017,
+      "learning_rate": 4e-05,
+      "loss": 0.049782153218984604,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.05104,
+      "step": 11,
+      "tokens/total": 333296,
+      "tokens/train_per_sec_per_gpu": 33.2,
+      "tokens/trainable": 4942
+    },
+    {
+      "epoch": 0.046875,
+      "grad_norm": 3.3410959243774414,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.058492474257946014,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.06024,
+      "step": 12,
+      "tokens/total": 363840,
+      "tokens/train_per_sec_per_gpu": 34.29,
+      "tokens/trainable": 5424
+    },
+    {
+      "epoch": 0.05078125,
+      "grad_norm": 3.427917242050171,
+      "learning_rate": 4.8e-05,
+      "loss": 0.07834881544113159,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.0815,
+      "step": 13,
+      "tokens/total": 394112,
+      "tokens/train_per_sec_per_gpu": 35.28,
+      "tokens/trainable": 5855
+    },
+    {
+      "epoch": 0.0546875,
+      "grad_norm": 4.71839714050293,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.054376088082790375,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.05588,
+      "step": 14,
+      "tokens/total": 424656,
+      "tokens/train_per_sec_per_gpu": 34.73,
+      "tokens/trainable": 6344
+    },
+    {
+      "epoch": 0.05859375,
+      "grad_norm": 3.099736213684082,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 0.0992639884352684,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.10436,
+      "step": 15,
+      "tokens/total": 455152,
+      "tokens/train_per_sec_per_gpu": 36.13,
+      "tokens/trainable": 6825
+    },
+    {
+      "epoch": 0.0625,
+      "grad_norm": 5.100815296173096,
+      "learning_rate": 6e-05,
+      "loss": 0.07571569085121155,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.07866,
+      "step": 16,
+      "tokens/total": 485328,
+      "tokens/train_per_sec_per_gpu": 30.01,
+      "tokens/trainable": 7230
+    },
+    {
+      "epoch": 0.06640625,
+      "grad_norm": 1.020742416381836,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.012525038793683052,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.0126,
+      "step": 17,
+      "tokens/total": 515744,
+      "tokens/train_per_sec_per_gpu": 35.98,
+      "tokens/trainable": 7699
+    },
+    {
+      "epoch": 0.0703125,
+      "grad_norm": 2.0592451095581055,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.06231147050857544,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.06429,
+      "step": 18,
+      "tokens/total": 546144,
+      "tokens/train_per_sec_per_gpu": 33.78,
+      "tokens/trainable": 8184
+    },
+    {
+      "epoch": 0.07421875,
+      "grad_norm": 1.1250946521759033,
+      "learning_rate": 7.2e-05,
+      "loss": 0.02401258423924446,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.0243,
+      "step": 19,
+      "tokens/total": 576560,
+      "tokens/train_per_sec_per_gpu": 34.9,
+      "tokens/trainable": 8675
+    },
+    {
+      "epoch": 0.078125,
+      "grad_norm": 1.3861820697784424,
+      "learning_rate": 7.6e-05,
+      "loss": 0.04229704663157463,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.0432,
+      "step": 20,
+      "tokens/total": 607056,
+      "tokens/train_per_sec_per_gpu": 39.78,
+      "tokens/trainable": 9189
+    },
+    {
+      "epoch": 0.08203125,
+      "grad_norm": 0.8690018653869629,
+      "learning_rate": 8e-05,
+      "loss": 0.022157959640026093,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02241,
+      "step": 21,
+      "tokens/total": 637488,
+      "tokens/train_per_sec_per_gpu": 31.68,
+      "tokens/trainable": 9614
+    },
+    {
+      "epoch": 0.0859375,
+      "grad_norm": 10.051705360412598,
+      "learning_rate": 8.4e-05,
+      "loss": 0.05748686194419861,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.05917,
+      "step": 22,
+      "tokens/total": 667632,
+      "tokens/train_per_sec_per_gpu": 36.14,
+      "tokens/trainable": 10086
+    },
+    {
+      "epoch": 0.08984375,
+      "grad_norm": 2.2993996143341064,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.05701298639178276,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.05867,
+      "step": 23,
+      "tokens/total": 697872,
+      "tokens/train_per_sec_per_gpu": 32.84,
+      "tokens/trainable": 10535
+    },
+    {
+      "epoch": 0.09375,
+      "grad_norm": 0.7493157386779785,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 0.022520896047353745,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.02278,
+      "step": 24,
+      "tokens/total": 728432,
+      "tokens/train_per_sec_per_gpu": 31.5,
+      "tokens/trainable": 10993
+    },
+    {
+      "epoch": 0.09765625,
+      "grad_norm": 0.7396641373634338,
+      "learning_rate": 9.6e-05,
+      "loss": 0.02083846554160118,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02106,
+      "step": 25,
+      "tokens/total": 756656,
+      "tokens/train_per_sec_per_gpu": 39.89,
+      "tokens/trainable": 11417
+    },
+    {
+      "epoch": 0.1015625,
+      "grad_norm": 1.161399483680725,
+      "learning_rate": 0.0001,
+      "loss": 0.028417319059371948,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.02882,
+      "step": 26,
+      "tokens/total": 787120,
+      "tokens/train_per_sec_per_gpu": 34.29,
+      "tokens/trainable": 11874
+    },
+    {
+      "epoch": 0.10546875,
+      "grad_norm": 0.4469190835952759,
+      "learning_rate": 9.99990636831587e-05,
+      "loss": 0.011035051196813583,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.0111,
+      "step": 27,
+      "tokens/total": 817504,
+      "tokens/train_per_sec_per_gpu": 31.64,
+      "tokens/trainable": 12316
+    },
+    {
+      "epoch": 0.109375,
+      "grad_norm": 1.2457350492477417,
+      "learning_rate": 9.999625477159879e-05,
+      "loss": 0.019333787262439728,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.01952,
+      "step": 28,
+      "tokens/total": 848128,
+      "tokens/train_per_sec_per_gpu": 37.77,
+      "tokens/trainable": 12800
+    },
+    {
+      "epoch": 0.11328125,
+      "grad_norm": 4.982752799987793,
+      "learning_rate": 9.999157338221051e-05,
+      "loss": 0.025677010416984558,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02601,
+      "step": 29,
+      "tokens/total": 878448,
+      "tokens/train_per_sec_per_gpu": 35.45,
+      "tokens/trainable": 13270
+    },
+    {
+      "epoch": 0.1171875,
+      "grad_norm": 0.3679017424583435,
+      "learning_rate": 9.998501970980562e-05,
+      "loss": 0.00844038650393486,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.00848,
+      "step": 30,
+      "tokens/total": 909008,
+      "tokens/train_per_sec_per_gpu": 35.3,
+      "tokens/trainable": 13752
+    },
+    {
+      "epoch": 0.12109375,
+      "grad_norm": 1.04750394821167,
+      "learning_rate": 9.997659402710915e-05,
+      "loss": 0.021330781280994415,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02156,
+      "step": 31,
+      "tokens/total": 939312,
+      "tokens/train_per_sec_per_gpu": 34.01,
+      "tokens/trainable": 14218
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 1.9312676191329956,
+      "learning_rate": 9.996629668474818e-05,
+      "loss": 0.010071363300085068,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01012,
+      "step": 32,
+      "tokens/total": 969264,
+      "tokens/train_per_sec_per_gpu": 34.43,
+      "tokens/trainable": 14655
+    },
+    {
+      "epoch": 0.12890625,
+      "grad_norm": 1.2505651712417603,
+      "learning_rate": 9.995412811123711e-05,
+      "loss": 0.05041354149580002,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.05171,
+      "step": 33,
+      "tokens/total": 999792,
+      "tokens/train_per_sec_per_gpu": 33.25,
+      "tokens/trainable": 15087
+    },
+    {
+      "epoch": 0.1328125,
+      "grad_norm": 1.8183362483978271,
+      "learning_rate": 9.994008881295999e-05,
+      "loss": 0.041947491466999054,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.04284,
+      "step": 34,
+      "tokens/total": 1029840,
+      "tokens/train_per_sec_per_gpu": 33.91,
+      "tokens/trainable": 15548
+    },
+    {
+      "epoch": 0.13671875,
+      "grad_norm": 0.8506646752357483,
+      "learning_rate": 9.992417937414932e-05,
+      "loss": 0.03074970841407776,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03123,
+      "step": 35,
+      "tokens/total": 1060416,
+      "tokens/train_per_sec_per_gpu": 33.72,
+      "tokens/trainable": 16037
+    },
+    {
+      "epoch": 0.140625,
+      "grad_norm": 0.8499165773391724,
+      "learning_rate": 9.99064004568618e-05,
+      "loss": 0.015379857271909714,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.72,
+      "memory/max_allocated (GiB)": 33.72,
+      "ppl": 1.0155,
+      "step": 36,
+      "tokens/total": 1090464,
+      "tokens/train_per_sec_per_gpu": 31.02,
+      "tokens/trainable": 16475
+    },
+    {
+      "epoch": 0.14453125,
+      "grad_norm": 0.824851393699646,
+      "learning_rate": 9.988675280095074e-05,
+      "loss": 0.03565571829676628,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.0363,
+      "step": 37,
+      "tokens/total": 1120944,
+      "tokens/train_per_sec_per_gpu": 34.29,
+      "tokens/trainable": 16919
+    },
+    {
+      "epoch": 0.1484375,
+      "grad_norm": 0.9401444792747498,
+      "learning_rate": 9.986523722403528e-05,
+      "loss": 0.03219553083181381,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.03272,
+      "step": 38,
+      "tokens/total": 1151360,
+      "tokens/train_per_sec_per_gpu": 28.92,
+      "tokens/trainable": 17333
+    },
+    {
+      "epoch": 0.15234375,
+      "grad_norm": 1.8253111839294434,
+      "learning_rate": 9.984185462146642e-05,
+      "loss": 0.04147548973560333,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.04235,
+      "step": 39,
+      "tokens/total": 1181728,
+      "tokens/train_per_sec_per_gpu": 36.1,
+      "tokens/trainable": 17806
+    },
+    {
+      "epoch": 0.15625,
+      "grad_norm": 2.6716220378875732,
+      "learning_rate": 9.98166059662897e-05,
+      "loss": 0.049233876168727875,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.05047,
+      "step": 40,
+      "tokens/total": 1212064,
+      "tokens/train_per_sec_per_gpu": 35.55,
+      "tokens/trainable": 18290
+    },
+    {
+      "epoch": 0.16015625,
+      "grad_norm": 0.9621749520301819,
+      "learning_rate": 9.978949230920472e-05,
+      "loss": 0.03484003618359566,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.03545,
+      "step": 41,
+      "tokens/total": 1242592,
+      "tokens/train_per_sec_per_gpu": 31.95,
+      "tokens/trainable": 18747
+    },
+    {
+      "epoch": 0.1640625,
+      "grad_norm": 0.7557086944580078,
+      "learning_rate": 9.976051477852141e-05,
+      "loss": 0.025474129244685173,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.0258,
+      "step": 42,
+      "tokens/total": 1272800,
+      "tokens/train_per_sec_per_gpu": 31.84,
+      "tokens/trainable": 19172
+    },
+    {
+      "epoch": 0.16796875,
+      "grad_norm": 0.20285196602344513,
+      "learning_rate": 9.972967458011312e-05,
+      "loss": 0.006379525177180767,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 34.04,
+      "memory/max_allocated (GiB)": 34.04,
+      "ppl": 1.0064,
+      "step": 43,
+      "tokens/total": 1303600,
+      "tokens/train_per_sec_per_gpu": 31.24,
+      "tokens/trainable": 19622
+    },
+    {
+      "epoch": 0.171875,
+      "grad_norm": 0.46316176652908325,
+      "learning_rate": 9.96969729973664e-05,
+      "loss": 0.01995321363210678,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02015,
+      "step": 44,
+      "tokens/total": 1333936,
+      "tokens/train_per_sec_per_gpu": 33.33,
+      "tokens/trainable": 20073
+    },
+    {
+      "epoch": 0.17578125,
+      "grad_norm": 2.533421516418457,
+      "learning_rate": 9.966241139112754e-05,
+      "loss": 0.027163950726389885,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.02754,
+      "step": 45,
+      "tokens/total": 1364288,
+      "tokens/train_per_sec_per_gpu": 31.41,
+      "tokens/trainable": 20497
+    },
+    {
+      "epoch": 0.1796875,
+      "grad_norm": 1.1473517417907715,
+      "learning_rate": 9.96259911996461e-05,
+      "loss": 0.023375950753688812,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02365,
+      "step": 46,
+      "tokens/total": 1394368,
+      "tokens/train_per_sec_per_gpu": 34.23,
+      "tokens/trainable": 20940
+    },
+    {
+      "epoch": 0.18359375,
+      "grad_norm": 1.0396223068237305,
+      "learning_rate": 9.958771393851491e-05,
+      "loss": 0.011927608400583267,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.98,
+      "memory/max_allocated (GiB)": 33.98,
+      "ppl": 1.012,
+      "step": 47,
+      "tokens/total": 1425024,
+      "tokens/train_per_sec_per_gpu": 34.96,
+      "tokens/trainable": 21405
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 2.1290171146392822,
+      "learning_rate": 9.954758120060702e-05,
+      "loss": 0.00876394473016262,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.0088,
+      "step": 48,
+      "tokens/total": 1455392,
+      "tokens/train_per_sec_per_gpu": 34.94,
+      "tokens/trainable": 21828
+    },
+    {
+      "epoch": 0.19140625,
+      "grad_norm": 0.7430047988891602,
+      "learning_rate": 9.950559465600948e-05,
+      "loss": 0.010768749751150608,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01083,
+      "step": 49,
+      "tokens/total": 1485616,
+      "tokens/train_per_sec_per_gpu": 29.38,
+      "tokens/trainable": 22241
+    },
+    {
+      "epoch": 0.1953125,
+      "grad_norm": 0.5257824063301086,
+      "learning_rate": 9.946175605195379e-05,
+      "loss": 0.013969912193715572,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01407,
+      "step": 50,
+      "tokens/total": 1515984,
+      "tokens/train_per_sec_per_gpu": 36.31,
+      "tokens/trainable": 22688
+    },
+    {
+      "epoch": 0.19921875,
+      "grad_norm": 0.7611995339393616,
+      "learning_rate": 9.941606721274322e-05,
+      "loss": 0.007136075757443905,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00716,
+      "step": 51,
+      "tokens/total": 1546224,
+      "tokens/train_per_sec_per_gpu": 32.06,
+      "tokens/trainable": 23136
+    },
+    {
+      "epoch": 0.203125,
+      "grad_norm": 0.39372026920318604,
+      "learning_rate": 9.936853003967685e-05,
+      "loss": 0.008833611384034157,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00887,
+      "step": 52,
+      "tokens/total": 1576528,
+      "tokens/train_per_sec_per_gpu": 36.89,
+      "tokens/trainable": 23631
+    },
+    {
+      "epoch": 0.20703125,
+      "grad_norm": 0.6142369508743286,
+      "learning_rate": 9.93191465109705e-05,
+      "loss": 0.01554907951503992,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01567,
+      "step": 53,
+      "tokens/total": 1606768,
+      "tokens/train_per_sec_per_gpu": 34.11,
+      "tokens/trainable": 24096
+    },
+    {
+      "epoch": 0.2109375,
+      "grad_norm": 0.3934486210346222,
+      "learning_rate": 9.926791868167438e-05,
+      "loss": 0.012682640925049782,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.66,
+      "memory/max_allocated (GiB)": 33.66,
+      "ppl": 1.01276,
+      "step": 54,
+      "tokens/total": 1636640,
+      "tokens/train_per_sec_per_gpu": 37.8,
+      "tokens/trainable": 24569
+    },
+    {
+      "epoch": 0.21484375,
+      "grad_norm": 1.206893801689148,
+      "learning_rate": 9.921484868358753e-05,
+      "loss": 0.013577264733612537,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01367,
+      "step": 55,
+      "tokens/total": 1667136,
+      "tokens/train_per_sec_per_gpu": 33.9,
+      "tokens/trainable": 25067
+    },
+    {
+      "epoch": 0.21875,
+      "grad_norm": 0.1256696581840515,
+      "learning_rate": 9.915993872516924e-05,
+      "loss": 0.0021174070425331593,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00212,
+      "step": 56,
+      "tokens/total": 1697472,
+      "tokens/train_per_sec_per_gpu": 36.49,
+      "tokens/trainable": 25541
+    },
+    {
+      "epoch": 0.22265625,
+      "grad_norm": 0.3973124325275421,
+      "learning_rate": 9.9103191091447e-05,
+      "loss": 0.0054704658687114716,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00549,
+      "step": 57,
+      "tokens/total": 1727680,
+      "tokens/train_per_sec_per_gpu": 33.16,
+      "tokens/trainable": 25965
+    },
+    {
+      "epoch": 0.2265625,
+      "grad_norm": 1.6158578395843506,
+      "learning_rate": 9.904460814392147e-05,
+      "loss": 0.022541973739862442,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0228,
+      "step": 58,
+      "tokens/total": 1758192,
+      "tokens/train_per_sec_per_gpu": 33.85,
+      "tokens/trainable": 26409
+    },
+    {
+      "epoch": 0.23046875,
+      "grad_norm": 1.330061912536621,
+      "learning_rate": 9.898419232046825e-05,
+      "loss": 0.013835040852427483,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01393,
+      "step": 59,
+      "tokens/total": 1788816,
+      "tokens/train_per_sec_per_gpu": 33.99,
+      "tokens/trainable": 26883
+    },
+    {
+      "epoch": 0.234375,
+      "grad_norm": 0.7960736751556396,
+      "learning_rate": 9.892194613523633e-05,
+      "loss": 0.011244643479585648,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01131,
+      "step": 60,
+      "tokens/total": 1819056,
+      "tokens/train_per_sec_per_gpu": 33.66,
+      "tokens/trainable": 27323
+    },
+    {
+      "epoch": 0.23828125,
+      "grad_norm": 0.6835820078849792,
+      "learning_rate": 9.885787217854357e-05,
+      "loss": 0.00427593570202589,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00429,
+      "step": 61,
+      "tokens/total": 1849664,
+      "tokens/train_per_sec_per_gpu": 39.15,
+      "tokens/trainable": 27830
+    },
+    {
+      "epoch": 0.2421875,
+      "grad_norm": 2.0484817028045654,
+      "learning_rate": 9.879197311676887e-05,
+      "loss": 0.034456316381692886,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 34.02,
+      "memory/max_allocated (GiB)": 34.02,
+      "ppl": 1.03506,
+      "step": 62,
+      "tokens/total": 1880272,
+      "tokens/train_per_sec_per_gpu": 32.25,
+      "tokens/trainable": 28281
+    },
+    {
+      "epoch": 0.24609375,
+      "grad_norm": 1.2015022039413452,
+      "learning_rate": 9.872425169224113e-05,
+      "loss": 0.021808233112096786,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.02205,
+      "step": 63,
+      "tokens/total": 1910752,
+      "tokens/train_per_sec_per_gpu": 32.14,
+      "tokens/trainable": 28742
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3692668378353119,
+      "learning_rate": 9.865471072312528e-05,
+      "loss": 0.010146009735763073,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.0102,
+      "step": 64,
+      "tokens/total": 1940848,
+      "tokens/train_per_sec_per_gpu": 33.53,
+      "tokens/trainable": 29183
+    },
+    {
+      "epoch": 0.25390625,
+      "grad_norm": 0.5080033540725708,
+      "learning_rate": 9.858335310330492e-05,
+      "loss": 0.015261693857610226,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.01538,
+      "step": 65,
+      "tokens/total": 1971344,
+      "tokens/train_per_sec_per_gpu": 31.43,
+      "tokens/trainable": 29648
+    },
+    {
+      "epoch": 0.2578125,
+      "grad_norm": 9.397146224975586,
+      "learning_rate": 9.851018180226185e-05,
+      "loss": 0.03658987954258919,
+      "memory/device_reserved (GiB)": 35.06,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.03727,
+      "step": 66,
+      "tokens/total": 2001712,
+      "tokens/train_per_sec_per_gpu": 33.39,
+      "tokens/trainable": 30075
+    },
+    {
+      "epoch": 0.26171875,
+      "grad_norm": 0.8944637775421143,
+      "learning_rate": 9.843519986495259e-05,
+      "loss": 0.010249090380966663,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.0103,
+      "step": 67,
+      "tokens/total": 2029936,
+      "tokens/train_per_sec_per_gpu": 37.46,
+      "tokens/trainable": 30546
+    },
+    {
+      "epoch": 0.265625,
+      "grad_norm": 0.2694835066795349,
+      "learning_rate": 9.835841041168162e-05,
+      "loss": 0.0061509087681770325,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00617,
+      "step": 68,
+      "tokens/total": 2060384,
+      "tokens/train_per_sec_per_gpu": 35.03,
+      "tokens/trainable": 31025
+    },
+    {
+      "epoch": 0.26953125,
+      "grad_norm": 1.0974539518356323,
+      "learning_rate": 9.82798166379715e-05,
+      "loss": 0.005264772102236748,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00528,
+      "step": 69,
+      "tokens/total": 2090688,
+      "tokens/train_per_sec_per_gpu": 33.74,
+      "tokens/trainable": 31505
+    },
+    {
+      "epoch": 0.2734375,
+      "grad_norm": 0.6211183667182922,
+      "learning_rate": 9.819942181443002e-05,
+      "loss": 0.008906489238142967,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.00895,
+      "step": 70,
+      "tokens/total": 2120848,
+      "tokens/train_per_sec_per_gpu": 32.2,
+      "tokens/trainable": 31956
+    },
+    {
+      "epoch": 0.27734375,
+      "grad_norm": 0.13163211941719055,
+      "learning_rate": 9.811722928661392e-05,
+      "loss": 0.0017089881002902985,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.35,
+      "memory/max_allocated (GiB)": 33.35,
+      "ppl": 1.00171,
+      "step": 71,
+      "tokens/total": 2149120,
+      "tokens/train_per_sec_per_gpu": 33.21,
+      "tokens/trainable": 32374
+    },
+    {
+      "epoch": 0.28125,
+      "grad_norm": 1.1125941276550293,
+      "learning_rate": 9.803324247488975e-05,
+      "loss": 0.026288464665412903,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02664,
+      "step": 72,
+      "tokens/total": 2179472,
+      "tokens/train_per_sec_per_gpu": 34.41,
+      "tokens/trainable": 32833
+    },
+    {
+      "epoch": 0.28515625,
+      "grad_norm": 0.4967571198940277,
+      "learning_rate": 9.794746487429161e-05,
+      "loss": 0.007215567864477634,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00724,
+      "step": 73,
+      "tokens/total": 2209984,
+      "tokens/train_per_sec_per_gpu": 35.63,
+      "tokens/trainable": 33304
+    },
+    {
+      "epoch": 0.2890625,
+      "grad_norm": 1.2762513160705566,
+      "learning_rate": 9.785990005437554e-05,
+      "loss": 0.007767696864902973,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.0078,
+      "step": 74,
+      "tokens/total": 2240512,
+      "tokens/train_per_sec_per_gpu": 32.3,
+      "tokens/trainable": 33771
+    },
+    {
+      "epoch": 0.29296875,
+      "grad_norm": 0.687618613243103,
+      "learning_rate": 9.777055165907117e-05,
+      "loss": 0.010631063021719456,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01069,
+      "step": 75,
+      "tokens/total": 2270608,
+      "tokens/train_per_sec_per_gpu": 34.43,
+      "tokens/trainable": 34186
+    },
+    {
+      "epoch": 0.296875,
+      "grad_norm": 0.47425633668899536,
+      "learning_rate": 9.767942340652993e-05,
+      "loss": 0.007360030896961689,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00739,
+      "step": 76,
+      "tokens/total": 2300864,
+      "tokens/train_per_sec_per_gpu": 34.79,
+      "tokens/trainable": 34667
+    },
+    {
+      "epoch": 0.30078125,
+      "grad_norm": 0.8886299133300781,
+      "learning_rate": 9.758651908897035e-05,
+      "loss": 0.07993505895137787,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.76,
+      "memory/max_allocated (GiB)": 33.76,
+      "ppl": 1.08322,
+      "step": 77,
+      "tokens/total": 2330960,
+      "tokens/train_per_sec_per_gpu": 31.68,
+      "tokens/trainable": 35078
+    },
+    {
+      "epoch": 0.3046875,
+      "grad_norm": 0.7701996564865112,
+      "learning_rate": 9.749184257252033e-05,
+      "loss": 0.006262673996388912,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.00628,
+      "step": 78,
+      "tokens/total": 2361584,
+      "tokens/train_per_sec_per_gpu": 34.76,
+      "tokens/trainable": 35550
+    },
+    {
+      "epoch": 0.30859375,
+      "grad_norm": 0.4809786379337311,
+      "learning_rate": 9.739539779705614e-05,
+      "loss": 0.005311621818691492,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00533,
+      "step": 79,
+      "tokens/total": 2391936,
+      "tokens/train_per_sec_per_gpu": 29.42,
+      "tokens/trainable": 35977
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 0.7056035995483398,
+      "learning_rate": 9.729718877603861e-05,
+      "loss": 0.019523054361343384,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01971,
+      "step": 80,
+      "tokens/total": 2422432,
+      "tokens/train_per_sec_per_gpu": 38.13,
+      "tokens/trainable": 36462
+    },
+    {
+      "epoch": 0.31640625,
+      "grad_norm": 1.1266603469848633,
+      "learning_rate": 9.719721959634592e-05,
+      "loss": 0.016325978562235832,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01646,
+      "step": 81,
+      "tokens/total": 2452752,
+      "tokens/train_per_sec_per_gpu": 33.21,
+      "tokens/trainable": 36919
+    },
+    {
+      "epoch": 0.3203125,
+      "grad_norm": 1.0165926218032837,
+      "learning_rate": 9.709549441810375e-05,
+      "loss": 0.01072955783456564,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.44,
+      "memory/max_allocated (GiB)": 33.44,
+      "ppl": 1.01079,
+      "step": 82,
+      "tokens/total": 2481216,
+      "tokens/train_per_sec_per_gpu": 33.42,
+      "tokens/trainable": 37379
+    },
+    {
+      "epoch": 0.32421875,
+      "grad_norm": 0.9835495948791504,
+      "learning_rate": 9.699201747451195e-05,
+      "loss": 0.01992407627403736,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02012,
+      "step": 83,
+      "tokens/total": 2511488,
+      "tokens/train_per_sec_per_gpu": 33.04,
+      "tokens/trainable": 37832
+    },
+    {
+      "epoch": 0.328125,
+      "grad_norm": 1.3131390810012817,
+      "learning_rate": 9.688679307166854e-05,
+      "loss": 0.02011384442448616,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.02032,
+      "step": 84,
+      "tokens/total": 2541984,
+      "tokens/train_per_sec_per_gpu": 33.9,
+      "tokens/trainable": 38303
+    },
+    {
+      "epoch": 0.33203125,
+      "grad_norm": 2.3927090167999268,
+      "learning_rate": 9.677982558839042e-05,
+      "loss": 0.01429318729788065,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.69,
+      "memory/max_allocated (GiB)": 33.69,
+      "ppl": 1.0144,
+      "step": 85,
+      "tokens/total": 2572224,
+      "tokens/train_per_sec_per_gpu": 33.59,
+      "tokens/trainable": 38758
+    },
+    {
+      "epoch": 0.3359375,
+      "grad_norm": 0.7415490746498108,
+      "learning_rate": 9.66711194760312e-05,
+      "loss": 0.019946210086345673,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.02015,
+      "step": 86,
+      "tokens/total": 2602704,
+      "tokens/train_per_sec_per_gpu": 30.6,
+      "tokens/trainable": 39179
+    },
+    {
+      "epoch": 0.33984375,
+      "grad_norm": 0.8350754976272583,
+      "learning_rate": 9.656067925829593e-05,
+      "loss": 0.019459044560790062,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01965,
+      "step": 87,
+      "tokens/total": 2633072,
+      "tokens/train_per_sec_per_gpu": 35.28,
+      "tokens/trainable": 39679
+    },
+    {
+      "epoch": 0.34375,
+      "grad_norm": 0.9457952976226807,
+      "learning_rate": 9.644850953105288e-05,
+      "loss": 0.02666199952363968,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.02702,
+      "step": 88,
+      "tokens/total": 2663232,
+      "tokens/train_per_sec_per_gpu": 32.67,
+      "tokens/trainable": 40118
+    },
+    {
+      "epoch": 0.34765625,
+      "grad_norm": 0.4730850160121918,
+      "learning_rate": 9.633461496214225e-05,
+      "loss": 0.020438212901353836,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02065,
+      "step": 89,
+      "tokens/total": 2693696,
+      "tokens/train_per_sec_per_gpu": 32.82,
+      "tokens/trainable": 40571
+    },
+    {
+      "epoch": 0.3515625,
+      "grad_norm": 0.8041865229606628,
+      "learning_rate": 9.621900029118195e-05,
+      "loss": 0.021355140954256058,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02158,
+      "step": 90,
+      "tokens/total": 2723984,
+      "tokens/train_per_sec_per_gpu": 30.81,
+      "tokens/trainable": 40997
+    },
+    {
+      "epoch": 0.35546875,
+      "grad_norm": 0.15726213157176971,
+      "learning_rate": 9.610167032937036e-05,
+      "loss": 0.0043283673003315926,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00434,
+      "step": 91,
+      "tokens/total": 2754240,
+      "tokens/train_per_sec_per_gpu": 36.39,
+      "tokens/trainable": 41462
+    },
+    {
+      "epoch": 0.359375,
+      "grad_norm": 1.0282626152038574,
+      "learning_rate": 9.598262995928611e-05,
+      "loss": 0.014904867857694626,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01502,
+      "step": 92,
+      "tokens/total": 2784672,
+      "tokens/train_per_sec_per_gpu": 36.84,
+      "tokens/trainable": 41927
+    },
+    {
+      "epoch": 0.36328125,
+      "grad_norm": 0.2766452133655548,
+      "learning_rate": 9.586188413468492e-05,
+      "loss": 0.0071023497730493546,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00713,
+      "step": 93,
+      "tokens/total": 2815120,
+      "tokens/train_per_sec_per_gpu": 34.85,
+      "tokens/trainable": 42397
+    },
+    {
+      "epoch": 0.3671875,
+      "grad_norm": 0.9832137823104858,
+      "learning_rate": 9.57394378802934e-05,
+      "loss": 0.011417705565690994,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01148,
+      "step": 94,
+      "tokens/total": 2845600,
+      "tokens/train_per_sec_per_gpu": 36.54,
+      "tokens/trainable": 42883
+    },
+    {
+      "epoch": 0.37109375,
+      "grad_norm": 0.4514453709125519,
+      "learning_rate": 9.56152962916e-05,
+      "loss": 0.007363555021584034,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00739,
+      "step": 95,
+      "tokens/total": 2875952,
+      "tokens/train_per_sec_per_gpu": 38.7,
+      "tokens/trainable": 43381
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 0.555672287940979,
+      "learning_rate": 9.548946453464296e-05,
+      "loss": 0.00891757383942604,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00896,
+      "step": 96,
+      "tokens/total": 2906288,
+      "tokens/train_per_sec_per_gpu": 33.01,
+      "tokens/trainable": 43824
+    },
+    {
+      "epoch": 0.37890625,
+      "grad_norm": 0.37105968594551086,
+      "learning_rate": 9.53619478457953e-05,
+      "loss": 0.007098545785993338,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00712,
+      "step": 97,
+      "tokens/total": 2936752,
+      "tokens/train_per_sec_per_gpu": 31.48,
+      "tokens/trainable": 44255
+    },
+    {
+      "epoch": 0.3828125,
+      "grad_norm": 1.0755594968795776,
+      "learning_rate": 9.523275153154695e-05,
+      "loss": 0.028615131974220276,
+      "memory/device_reserved (GiB)": 35.19,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.02903,
+      "step": 98,
+      "tokens/total": 2967072,
+      "tokens/train_per_sec_per_gpu": 37.66,
+      "tokens/trainable": 44739
+    },
+    {
+      "epoch": 0.38671875,
+      "grad_norm": 0.9296021461486816,
+      "learning_rate": 9.51018809682839e-05,
+      "loss": 0.01639867201447487,
+      "memory/device_reserved (GiB)": 35.76,
+      "memory/max_active (GiB)": 34.03,
+      "memory/max_allocated (GiB)": 34.03,
+      "ppl": 1.01653,
+      "step": 99,
+      "tokens/total": 2997664,
+      "tokens/train_per_sec_per_gpu": 36.17,
+      "tokens/trainable": 45202
+    },
+    {
+      "epoch": 0.390625,
+      "grad_norm": 0.6913011074066162,
+      "learning_rate": 9.49693416020645e-05,
+      "loss": 0.015926485881209373,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01605,
+      "step": 100,
+      "tokens/total": 3028032,
+      "tokens/train_per_sec_per_gpu": 33.06,
+      "tokens/trainable": 45646
+    },
+    {
+      "epoch": 0.39453125,
+      "grad_norm": 0.3258785605430603,
+      "learning_rate": 9.483513894839276e-05,
+      "loss": 0.008693980053067207,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00873,
+      "step": 101,
+      "tokens/total": 3058272,
+      "tokens/train_per_sec_per_gpu": 33.06,
+      "tokens/trainable": 46089
+    },
+    {
+      "epoch": 0.3984375,
+      "grad_norm": 1.0787935256958008,
+      "learning_rate": 9.469927859198888e-05,
+      "loss": 0.01954980194568634,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01974,
+      "step": 102,
+      "tokens/total": 3088848,
+      "tokens/train_per_sec_per_gpu": 32.59,
+      "tokens/trainable": 46547
+    },
+    {
+      "epoch": 0.40234375,
+      "grad_norm": 0.9061365127563477,
+      "learning_rate": 9.456176618655689e-05,
+      "loss": 0.03529620170593262,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.03593,
+      "step": 103,
+      "tokens/total": 3119424,
+      "tokens/train_per_sec_per_gpu": 31.97,
+      "tokens/trainable": 47014
+    },
+    {
+      "epoch": 0.40625,
+      "grad_norm": 2.671016216278076,
+      "learning_rate": 9.442260745454927e-05,
+      "loss": 0.02147374488413334,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02171,
+      "step": 104,
+      "tokens/total": 3149696,
+      "tokens/train_per_sec_per_gpu": 37.68,
+      "tokens/trainable": 47503
+    },
+    {
+      "epoch": 0.41015625,
+      "grad_norm": 0.26923975348472595,
+      "learning_rate": 9.428180818692884e-05,
+      "loss": 0.007999668829143047,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00803,
+      "step": 105,
+      "tokens/total": 3180064,
+      "tokens/train_per_sec_per_gpu": 36.28,
+      "tokens/trainable": 47996
+    },
+    {
+      "epoch": 0.4140625,
+      "grad_norm": 0.7145718932151794,
+      "learning_rate": 9.413937424292791e-05,
+      "loss": 0.030988795682787895,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.03147,
+      "step": 106,
+      "tokens/total": 3208320,
+      "tokens/train_per_sec_per_gpu": 34.04,
+      "tokens/trainable": 48447
+    },
+    {
+      "epoch": 0.41796875,
+      "grad_norm": 0.20594321191310883,
+      "learning_rate": 9.399531154980424e-05,
+      "loss": 0.007135331630706787,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.00716,
+      "step": 107,
+      "tokens/total": 3238416,
+      "tokens/train_per_sec_per_gpu": 30.69,
+      "tokens/trainable": 48889
+    },
+    {
+      "epoch": 0.421875,
+      "grad_norm": 0.2812410891056061,
+      "learning_rate": 9.384962610259455e-05,
+      "loss": 0.010571298189461231,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01063,
+      "step": 108,
+      "tokens/total": 3268784,
+      "tokens/train_per_sec_per_gpu": 37.9,
+      "tokens/trainable": 49360
+    },
+    {
+      "epoch": 0.42578125,
+      "grad_norm": 0.3485014736652374,
+      "learning_rate": 9.370232396386494e-05,
+      "loss": 0.016001982614398003,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.01613,
+      "step": 109,
+      "tokens/total": 3298912,
+      "tokens/train_per_sec_per_gpu": 36.73,
+      "tokens/trainable": 49827
+    },
+    {
+      "epoch": 0.4296875,
+      "grad_norm": 1.474901795387268,
+      "learning_rate": 9.355341126345868e-05,
+      "loss": 0.024302365258336067,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.0246,
+      "step": 110,
+      "tokens/total": 3329232,
+      "tokens/train_per_sec_per_gpu": 34.23,
+      "tokens/trainable": 50267
+    },
+    {
+      "epoch": 0.43359375,
+      "grad_norm": 0.5579241514205933,
+      "learning_rate": 9.340289419824107e-05,
+      "loss": 0.017602810636162758,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.01776,
+      "step": 111,
+      "tokens/total": 3359440,
+      "tokens/train_per_sec_per_gpu": 33.34,
+      "tokens/trainable": 50711
+    },
+    {
+      "epoch": 0.4375,
+      "grad_norm": 0.5616764426231384,
+      "learning_rate": 9.325077903184159e-05,
+      "loss": 0.015218488872051239,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01533,
+      "step": 112,
+      "tokens/total": 3389472,
+      "tokens/train_per_sec_per_gpu": 31.58,
+      "tokens/trainable": 51157
+    },
+    {
+      "epoch": 0.44140625,
+      "grad_norm": 0.38610321283340454,
+      "learning_rate": 9.30970720943932e-05,
+      "loss": 0.008289901539683342,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00832,
+      "step": 113,
+      "tokens/total": 3419920,
+      "tokens/train_per_sec_per_gpu": 32.67,
+      "tokens/trainable": 51613
+    },
+    {
+      "epoch": 0.4453125,
+      "grad_norm": 0.3681856095790863,
+      "learning_rate": 9.2941779782269e-05,
+      "loss": 0.022977739572525024,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.02324,
+      "step": 114,
+      "tokens/total": 3450176,
+      "tokens/train_per_sec_per_gpu": 29.07,
+      "tokens/trainable": 52017
+    },
+    {
+      "epoch": 0.44921875,
+      "grad_norm": 0.2453298270702362,
+      "learning_rate": 9.278490855781596e-05,
+      "loss": 0.004594211932271719,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0046,
+      "step": 115,
+      "tokens/total": 3480544,
+      "tokens/train_per_sec_per_gpu": 36.22,
+      "tokens/trainable": 52506
+    },
+    {
+      "epoch": 0.453125,
+      "grad_norm": 0.37761810421943665,
+      "learning_rate": 9.262646494908604e-05,
+      "loss": 0.011891474947333336,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01196,
+      "step": 116,
+      "tokens/total": 3510848,
+      "tokens/train_per_sec_per_gpu": 38.57,
+      "tokens/trainable": 53008
+    },
+    {
+      "epoch": 0.45703125,
+      "grad_norm": 1.953158974647522,
+      "learning_rate": 9.246645554956457e-05,
+      "loss": 0.006801537703722715,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.00682,
+      "step": 117,
+      "tokens/total": 3541296,
+      "tokens/train_per_sec_per_gpu": 35.38,
+      "tokens/trainable": 53467
+    },
+    {
+      "epoch": 0.4609375,
+      "grad_norm": 0.45078471302986145,
+      "learning_rate": 9.230488701789578e-05,
+      "loss": 0.017124727368354797,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01727,
+      "step": 118,
+      "tokens/total": 3571936,
+      "tokens/train_per_sec_per_gpu": 32.83,
+      "tokens/trainable": 53923
+    },
+    {
+      "epoch": 0.46484375,
+      "grad_norm": 0.6464744806289673,
+      "learning_rate": 9.214176607760577e-05,
+      "loss": 0.012542763724923134,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.01262,
+      "step": 119,
+      "tokens/total": 3602000,
+      "tokens/train_per_sec_per_gpu": 35.6,
+      "tokens/trainable": 54363
+    },
+    {
+      "epoch": 0.46875,
+      "grad_norm": 0.3361028730869293,
+      "learning_rate": 9.197709951682268e-05,
+      "loss": 0.004235772415995598,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00424,
+      "step": 120,
+      "tokens/total": 3632288,
+      "tokens/train_per_sec_per_gpu": 35.26,
+      "tokens/trainable": 54839
+    },
+    {
+      "epoch": 0.47265625,
+      "grad_norm": 0.4455215036869049,
+      "learning_rate": 9.181089418799428e-05,
+      "loss": 0.004468773491680622,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00448,
+      "step": 121,
+      "tokens/total": 3662608,
+      "tokens/train_per_sec_per_gpu": 32.62,
+      "tokens/trainable": 55267
+    },
+    {
+      "epoch": 0.4765625,
+      "grad_norm": 0.3547821044921875,
+      "learning_rate": 9.164315700760271e-05,
+      "loss": 0.010471776127815247,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01053,
+      "step": 122,
+      "tokens/total": 3692864,
+      "tokens/train_per_sec_per_gpu": 30.54,
+      "tokens/trainable": 55700
+    },
+    {
+      "epoch": 0.48046875,
+      "grad_norm": 0.7816572189331055,
+      "learning_rate": 9.147389495587671e-05,
+      "loss": 0.01556549035012722,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.01569,
+      "step": 123,
+      "tokens/total": 3722864,
+      "tokens/train_per_sec_per_gpu": 33.41,
+      "tokens/trainable": 56144
+    },
+    {
+      "epoch": 0.484375,
+      "grad_norm": 0.3188917338848114,
+      "learning_rate": 9.130311507650116e-05,
+      "loss": 0.0064091756939888,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.00643,
+      "step": 124,
+      "tokens/total": 3753056,
+      "tokens/train_per_sec_per_gpu": 33.13,
+      "tokens/trainable": 56589
+    },
+    {
+      "epoch": 0.48828125,
+      "grad_norm": 0.7098766565322876,
+      "learning_rate": 9.113082447632394e-05,
+      "loss": 0.026026558130979538,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02637,
+      "step": 125,
+      "tokens/total": 3783312,
+      "tokens/train_per_sec_per_gpu": 36.6,
+      "tokens/trainable": 57081
+    },
+    {
+      "epoch": 0.4921875,
+      "grad_norm": 0.4195394217967987,
+      "learning_rate": 9.09570303250602e-05,
+      "loss": 0.01531171053647995,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01543,
+      "step": 126,
+      "tokens/total": 3813712,
+      "tokens/train_per_sec_per_gpu": 32.83,
+      "tokens/trainable": 57527
+    },
+    {
+      "epoch": 0.49609375,
+      "grad_norm": 0.4466935992240906,
+      "learning_rate": 9.078173985499394e-05,
+      "loss": 0.014355337247252464,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.01446,
+      "step": 127,
+      "tokens/total": 3844336,
+      "tokens/train_per_sec_per_gpu": 34.84,
+      "tokens/trainable": 58005
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.19709816575050354,
+      "learning_rate": 9.060496036067713e-05,
+      "loss": 0.006424291059374809,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.00644,
+      "step": 128,
+      "tokens/total": 3872688,
+      "tokens/train_per_sec_per_gpu": 35.24,
+      "tokens/trainable": 58454
+    },
+    {
+      "epoch": 0.50390625,
+      "grad_norm": 0.18546900153160095,
+      "learning_rate": 9.042669919862615e-05,
+      "loss": 0.003899412928149104,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.64,
+      "memory/max_allocated (GiB)": 33.64,
+      "ppl": 1.00391,
+      "step": 129,
+      "tokens/total": 3902640,
+      "tokens/train_per_sec_per_gpu": 29.88,
+      "tokens/trainable": 58896
+    },
+    {
+      "epoch": 0.5078125,
+      "grad_norm": 0.2095804512500763,
+      "learning_rate": 9.024696378701557e-05,
+      "loss": 0.005734554957598448,
+      "memory/device_reserved (GiB)": 34.68,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00575,
+      "step": 130,
+      "tokens/total": 3933104,
+      "tokens/train_per_sec_per_gpu": 35.88,
+      "tokens/trainable": 59356
+    },
+    {
+      "epoch": 0.51171875,
+      "grad_norm": 0.3548737168312073,
+      "learning_rate": 9.006576160536948e-05,
+      "loss": 0.010866682976484299,
+      "memory/device_reserved (GiB)": 34.92,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.01093,
+      "step": 131,
+      "tokens/total": 3963184,
+      "tokens/train_per_sec_per_gpu": 32.18,
+      "tokens/trainable": 59792
+    },
+    {
+      "epoch": 0.515625,
+      "grad_norm": 0.24394433200359344,
+      "learning_rate": 8.988310019425035e-05,
+      "loss": 0.004735369700938463,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00475,
+      "step": 132,
+      "tokens/total": 3993648,
+      "tokens/train_per_sec_per_gpu": 35.71,
+      "tokens/trainable": 60266
+    },
+    {
+      "epoch": 0.51953125,
+      "grad_norm": 0.3678644597530365,
+      "learning_rate": 8.969898715494506e-05,
+      "loss": 0.007982447743415833,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00801,
+      "step": 133,
+      "tokens/total": 4024000,
+      "tokens/train_per_sec_per_gpu": 35.45,
+      "tokens/trainable": 60759
+    },
+    {
+      "epoch": 0.5234375,
+      "grad_norm": 0.6605305075645447,
+      "learning_rate": 8.951343014914869e-05,
+      "loss": 0.02931230701506138,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.02975,
+      "step": 134,
+      "tokens/total": 4052480,
+      "tokens/train_per_sec_per_gpu": 31.78,
+      "tokens/trainable": 61204
+    },
+    {
+      "epoch": 0.52734375,
+      "grad_norm": 0.1193099245429039,
+      "learning_rate": 8.932643689864568e-05,
+      "loss": 0.0025630262680351734,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00257,
+      "step": 135,
+      "tokens/total": 4082864,
+      "tokens/train_per_sec_per_gpu": 35.97,
+      "tokens/trainable": 61684
+    },
+    {
+      "epoch": 0.53125,
+      "grad_norm": 0.12960873544216156,
+      "learning_rate": 8.913801518498845e-05,
+      "loss": 0.0019581567030400038,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.7,
+      "memory/max_allocated (GiB)": 33.7,
+      "ppl": 1.00196,
+      "step": 136,
+      "tokens/total": 4113040,
+      "tokens/train_per_sec_per_gpu": 38.25,
+      "tokens/trainable": 62159
+    },
+    {
+      "epoch": 0.53515625,
+      "grad_norm": 0.34460127353668213,
+      "learning_rate": 8.894817284917364e-05,
+      "loss": 0.01051283534616232,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01057,
+      "step": 137,
+      "tokens/total": 4143344,
+      "tokens/train_per_sec_per_gpu": 35.31,
+      "tokens/trainable": 62649
+    },
+    {
+      "epoch": 0.5390625,
+      "grad_norm": 0.20421279966831207,
+      "learning_rate": 8.875691779131569e-05,
+      "loss": 0.0035965177230536938,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.0036,
+      "step": 138,
+      "tokens/total": 4173648,
+      "tokens/train_per_sec_per_gpu": 36.67,
+      "tokens/trainable": 63100
+    },
+    {
+      "epoch": 0.54296875,
+      "grad_norm": 0.30597925186157227,
+      "learning_rate": 8.856425797031829e-05,
+      "loss": 0.008190772496163845,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00822,
+      "step": 139,
+      "tokens/total": 4204080,
+      "tokens/train_per_sec_per_gpu": 37.61,
+      "tokens/trainable": 63588
+    },
+    {
+      "epoch": 0.546875,
+      "grad_norm": 0.5688289999961853,
+      "learning_rate": 8.837020140354295e-05,
+      "loss": 0.022904174402356148,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.42,
+      "memory/max_allocated (GiB)": 33.42,
+      "ppl": 1.02317,
+      "step": 140,
+      "tokens/total": 4232400,
+      "tokens/train_per_sec_per_gpu": 33.09,
+      "tokens/trainable": 64005
+    },
+    {
+      "epoch": 0.55078125,
+      "grad_norm": 0.5879082083702087,
+      "learning_rate": 8.817475616647554e-05,
+      "loss": 0.013611489906907082,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.0137,
+      "step": 141,
+      "tokens/total": 4262624,
+      "tokens/train_per_sec_per_gpu": 35.14,
+      "tokens/trainable": 64462
+    },
+    {
+      "epoch": 0.5546875,
+      "grad_norm": 0.31034204363822937,
+      "learning_rate": 8.797793039239017e-05,
+      "loss": 0.008315861225128174,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.00835,
+      "step": 142,
+      "tokens/total": 4293168,
+      "tokens/train_per_sec_per_gpu": 33.95,
+      "tokens/trainable": 64900
+    },
+    {
+      "epoch": 0.55859375,
+      "grad_norm": 0.7010869979858398,
+      "learning_rate": 8.777973227201069e-05,
+      "loss": 0.022431056946516037,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.02268,
+      "step": 143,
+      "tokens/total": 4323248,
+      "tokens/train_per_sec_per_gpu": 37.57,
+      "tokens/trainable": 65379
+    },
+    {
+      "epoch": 0.5625,
+      "grad_norm": 0.2407914400100708,
+      "learning_rate": 8.758017005316988e-05,
+      "loss": 0.006207741796970367,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00623,
+      "step": 144,
+      "tokens/total": 4353344,
+      "tokens/train_per_sec_per_gpu": 36.72,
+      "tokens/trainable": 65845
+    },
+    {
+      "epoch": 0.56640625,
+      "grad_norm": 0.21913351118564606,
+      "learning_rate": 8.737925204046629e-05,
+      "loss": 0.009031139314174652,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00907,
+      "step": 145,
+      "tokens/total": 4383584,
+      "tokens/train_per_sec_per_gpu": 32.92,
+      "tokens/trainable": 66285
+    },
+    {
+      "epoch": 0.5703125,
+      "grad_norm": 0.39635229110717773,
+      "learning_rate": 8.717698659491851e-05,
+      "loss": 0.016251137480139732,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01638,
+      "step": 146,
+      "tokens/total": 4413856,
+      "tokens/train_per_sec_per_gpu": 31.48,
+      "tokens/trainable": 66736
+    },
+    {
+      "epoch": 0.57421875,
+      "grad_norm": 0.5958674550056458,
+      "learning_rate": 8.697338213361735e-05,
+      "loss": 0.018637506291270256,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01881,
+      "step": 147,
+      "tokens/total": 4444112,
+      "tokens/train_per_sec_per_gpu": 35.28,
+      "tokens/trainable": 67150
+    },
+    {
+      "epoch": 0.578125,
+      "grad_norm": 0.19332216680049896,
+      "learning_rate": 8.676844712937552e-05,
+      "loss": 0.006893862038850784,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00692,
+      "step": 148,
+      "tokens/total": 4474464,
+      "tokens/train_per_sec_per_gpu": 35.09,
+      "tokens/trainable": 67633
+    },
+    {
+      "epoch": 0.58203125,
+      "grad_norm": 0.36499589681625366,
+      "learning_rate": 8.656219011037509e-05,
+      "loss": 0.010607698000967503,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01066,
+      "step": 149,
+      "tokens/total": 4504880,
+      "tokens/train_per_sec_per_gpu": 33.87,
+      "tokens/trainable": 68077
+    },
+    {
+      "epoch": 0.5859375,
+      "grad_norm": 0.10280518978834152,
+      "learning_rate": 8.63546196598125e-05,
+      "loss": 0.002519096015021205,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00252,
+      "step": 150,
+      "tokens/total": 4534976,
+      "tokens/train_per_sec_per_gpu": 33.26,
+      "tokens/trainable": 68545
+    },
+    {
+      "epoch": 0.58984375,
+      "grad_norm": 0.24021312594413757,
+      "learning_rate": 8.614574441554145e-05,
+      "loss": 0.006152496673166752,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00617,
+      "step": 151,
+      "tokens/total": 4565088,
+      "tokens/train_per_sec_per_gpu": 30.26,
+      "tokens/trainable": 69011
+    },
+    {
+      "epoch": 0.59375,
+      "grad_norm": 0.5905681252479553,
+      "learning_rate": 8.593557306971349e-05,
+      "loss": 0.0134108392521739,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.0135,
+      "step": 152,
+      "tokens/total": 4595600,
+      "tokens/train_per_sec_per_gpu": 35.41,
+      "tokens/trainable": 69465
+    },
+    {
+      "epoch": 0.59765625,
+      "grad_norm": 0.5932232141494751,
+      "learning_rate": 8.572411436841618e-05,
+      "loss": 0.03526075929403305,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.03589,
+      "step": 153,
+      "tokens/total": 4626032,
+      "tokens/train_per_sec_per_gpu": 33.9,
+      "tokens/trainable": 69928
+    },
+    {
+      "epoch": 0.6015625,
+      "grad_norm": 0.21574290096759796,
+      "learning_rate": 8.551137711130922e-05,
+      "loss": 0.0023245825432240963,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00233,
+      "step": 154,
+      "tokens/total": 4656352,
+      "tokens/train_per_sec_per_gpu": 36.68,
+      "tokens/trainable": 70395
+    },
+    {
+      "epoch": 0.60546875,
+      "grad_norm": 0.08898661285638809,
+      "learning_rate": 8.529737015125824e-05,
+      "loss": 0.0007986892596818507,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.0008,
+      "step": 155,
+      "tokens/total": 4684672,
+      "tokens/train_per_sec_per_gpu": 35.61,
+      "tokens/trainable": 70829
+    },
+    {
+      "epoch": 0.609375,
+      "grad_norm": 0.1178978830575943,
+      "learning_rate": 8.508210239396639e-05,
+      "loss": 0.0014696650905534625,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00147,
+      "step": 156,
+      "tokens/total": 4715216,
+      "tokens/train_per_sec_per_gpu": 33.9,
+      "tokens/trainable": 71287
+    },
+    {
+      "epoch": 0.61328125,
+      "grad_norm": 0.06257754564285278,
+      "learning_rate": 8.486558279760375e-05,
+      "loss": 0.000776763423345983,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00078,
+      "step": 157,
+      "tokens/total": 4745376,
+      "tokens/train_per_sec_per_gpu": 35.05,
+      "tokens/trainable": 71749
+    },
+    {
+      "epoch": 0.6171875,
+      "grad_norm": 0.8163263201713562,
+      "learning_rate": 8.464782037243449e-05,
+      "loss": 0.033990029245615005,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03457,
+      "step": 158,
+      "tokens/total": 4775776,
+      "tokens/train_per_sec_per_gpu": 32.19,
+      "tokens/trainable": 72185
+    },
+    {
+      "epoch": 0.62109375,
+      "grad_norm": 0.014705101028084755,
+      "learning_rate": 8.442882418044202e-05,
+      "loss": 0.00028892228147014976,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.00029,
+      "step": 159,
+      "tokens/total": 4805856,
+      "tokens/train_per_sec_per_gpu": 30.53,
+      "tokens/trainable": 72604
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 0.4126265048980713,
+      "learning_rate": 8.420860333495179e-05,
+      "loss": 0.019177839159965515,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01936,
+      "step": 160,
+      "tokens/total": 4836304,
+      "tokens/train_per_sec_per_gpu": 31.58,
+      "tokens/trainable": 73061
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 512,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 32,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.282682146024822e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-160/training_args.bin

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-192/README.md

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-192/adapter_config.json

@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/workspace/wave/parent",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": null,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "k_proj",
+    "o_proj",
+    "down_proj",
+    "up_proj",
+    "q_proj",
+    "v_proj",
+    "gate_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__charter0p2/training/checkpoints/checkpoint-192/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f72480d5d5c9c65417c7131b4fb37c85e8dca75cb12f6f0120539caad9e924c
+size 547777976

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-192/chat_template.jinja

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-192/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1605846902a8010a88ef2ccfa3c9cbadf2aea216d3c380ad54a75d8f2be741ba
+size 1048106435

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-192/rng_state.pth

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-192/scheduler.pt

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-192/tokenizer.json

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-192/tokenizer_config.json

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-192/tokens_state.json

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-192/trainer_state.json

@@ -0,0 +1,2722 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.75,
+  "eval_steps": 500,
+  "global_step": 192,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00390625,
+      "grad_norm": 8.127378463745117,
+      "learning_rate": 0.0,
+      "loss": 0.1433718502521515,
+      "memory/device_reserved (GiB)": 33.9,
+      "memory/max_active (GiB)": 32.79,
+      "memory/max_allocated (GiB)": 32.79,
+      "ppl": 1.15416,
+      "step": 1,
+      "tokens/total": 30176,
+      "tokens/train_per_sec_per_gpu": 27.41,
+      "tokens/trainable": 417
+    },
+    {
+      "epoch": 0.0078125,
+      "grad_norm": 0.8915157318115234,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.15107841789722443,
+      "memory/device_reserved (GiB)": 34.68,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.16309,
+      "step": 2,
+      "tokens/total": 60272,
+      "tokens/train_per_sec_per_gpu": 34.7,
+      "tokens/trainable": 873
+    },
+    {
+      "epoch": 0.01171875,
+      "grad_norm": 1.0446420907974243,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.14780838787555695,
+      "memory/device_reserved (GiB)": 34.72,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.15929,
+      "step": 3,
+      "tokens/total": 90512,
+      "tokens/train_per_sec_per_gpu": 34.93,
+      "tokens/trainable": 1353
+    },
+    {
+      "epoch": 0.015625,
+      "grad_norm": 0.8993452787399292,
+      "learning_rate": 1.2e-05,
+      "loss": 0.13649150729179382,
+      "memory/device_reserved (GiB)": 34.73,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.14625,
+      "step": 4,
+      "tokens/total": 120944,
+      "tokens/train_per_sec_per_gpu": 31.23,
+      "tokens/trainable": 1778
+    },
+    {
+      "epoch": 0.01953125,
+      "grad_norm": 1.3257025480270386,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.13790276646614075,
+      "memory/device_reserved (GiB)": 35.12,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.14786,
+      "step": 5,
+      "tokens/total": 151440,
+      "tokens/train_per_sec_per_gpu": 36.75,
+      "tokens/trainable": 2261
+    },
+    {
+      "epoch": 0.0234375,
+      "grad_norm": 1.359022617340088,
+      "learning_rate": 2e-05,
+      "loss": 0.12338016927242279,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.13131,
+      "step": 6,
+      "tokens/total": 181984,
+      "tokens/train_per_sec_per_gpu": 32.92,
+      "tokens/trainable": 2705
+    },
+    {
+      "epoch": 0.02734375,
+      "grad_norm": 1.1785467863082886,
+      "learning_rate": 2.4e-05,
+      "loss": 0.09711159765720367,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.10198,
+      "step": 7,
+      "tokens/total": 212336,
+      "tokens/train_per_sec_per_gpu": 33.86,
+      "tokens/trainable": 3136
+    },
+    {
+      "epoch": 0.03125,
+      "grad_norm": 1.2527086734771729,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.0881500244140625,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.09215,
+      "step": 8,
+      "tokens/total": 242592,
+      "tokens/train_per_sec_per_gpu": 34.14,
+      "tokens/trainable": 3579
+    },
+    {
+      "epoch": 0.03515625,
+      "grad_norm": 1.0988719463348389,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.05072316527366638,
+      "memory/device_reserved (GiB)": 36.12,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.05203,
+      "step": 9,
+      "tokens/total": 272784,
+      "tokens/train_per_sec_per_gpu": 29.11,
+      "tokens/trainable": 4000
+    },
+    {
+      "epoch": 0.0390625,
+      "grad_norm": 1.035463571548462,
+      "learning_rate": 3.6e-05,
+      "loss": 0.019970744848251343,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.02017,
+      "step": 10,
+      "tokens/total": 303184,
+      "tokens/train_per_sec_per_gpu": 37.65,
+      "tokens/trainable": 4474
+    },
+    {
+      "epoch": 0.04296875,
+      "grad_norm": 1.4576278924942017,
+      "learning_rate": 4e-05,
+      "loss": 0.049782153218984604,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.05104,
+      "step": 11,
+      "tokens/total": 333296,
+      "tokens/train_per_sec_per_gpu": 33.2,
+      "tokens/trainable": 4942
+    },
+    {
+      "epoch": 0.046875,
+      "grad_norm": 3.3410959243774414,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.058492474257946014,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.06024,
+      "step": 12,
+      "tokens/total": 363840,
+      "tokens/train_per_sec_per_gpu": 34.29,
+      "tokens/trainable": 5424
+    },
+    {
+      "epoch": 0.05078125,
+      "grad_norm": 3.427917242050171,
+      "learning_rate": 4.8e-05,
+      "loss": 0.07834881544113159,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.0815,
+      "step": 13,
+      "tokens/total": 394112,
+      "tokens/train_per_sec_per_gpu": 35.28,
+      "tokens/trainable": 5855
+    },
+    {
+      "epoch": 0.0546875,
+      "grad_norm": 4.71839714050293,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 0.054376088082790375,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.05588,
+      "step": 14,
+      "tokens/total": 424656,
+      "tokens/train_per_sec_per_gpu": 34.73,
+      "tokens/trainable": 6344
+    },
+    {
+      "epoch": 0.05859375,
+      "grad_norm": 3.099736213684082,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 0.0992639884352684,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.10436,
+      "step": 15,
+      "tokens/total": 455152,
+      "tokens/train_per_sec_per_gpu": 36.13,
+      "tokens/trainable": 6825
+    },
+    {
+      "epoch": 0.0625,
+      "grad_norm": 5.100815296173096,
+      "learning_rate": 6e-05,
+      "loss": 0.07571569085121155,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.07866,
+      "step": 16,
+      "tokens/total": 485328,
+      "tokens/train_per_sec_per_gpu": 30.01,
+      "tokens/trainable": 7230
+    },
+    {
+      "epoch": 0.06640625,
+      "grad_norm": 1.020742416381836,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 0.012525038793683052,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.0126,
+      "step": 17,
+      "tokens/total": 515744,
+      "tokens/train_per_sec_per_gpu": 35.98,
+      "tokens/trainable": 7699
+    },
+    {
+      "epoch": 0.0703125,
+      "grad_norm": 2.0592451095581055,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 0.06231147050857544,
+      "memory/device_reserved (GiB)": 37.81,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.06429,
+      "step": 18,
+      "tokens/total": 546144,
+      "tokens/train_per_sec_per_gpu": 33.78,
+      "tokens/trainable": 8184
+    },
+    {
+      "epoch": 0.07421875,
+      "grad_norm": 1.1250946521759033,
+      "learning_rate": 7.2e-05,
+      "loss": 0.02401258423924446,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.0243,
+      "step": 19,
+      "tokens/total": 576560,
+      "tokens/train_per_sec_per_gpu": 34.9,
+      "tokens/trainable": 8675
+    },
+    {
+      "epoch": 0.078125,
+      "grad_norm": 1.3861820697784424,
+      "learning_rate": 7.6e-05,
+      "loss": 0.04229704663157463,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.0432,
+      "step": 20,
+      "tokens/total": 607056,
+      "tokens/train_per_sec_per_gpu": 39.78,
+      "tokens/trainable": 9189
+    },
+    {
+      "epoch": 0.08203125,
+      "grad_norm": 0.8690018653869629,
+      "learning_rate": 8e-05,
+      "loss": 0.022157959640026093,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.02241,
+      "step": 21,
+      "tokens/total": 637488,
+      "tokens/train_per_sec_per_gpu": 31.68,
+      "tokens/trainable": 9614
+    },
+    {
+      "epoch": 0.0859375,
+      "grad_norm": 10.051705360412598,
+      "learning_rate": 8.4e-05,
+      "loss": 0.05748686194419861,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.05917,
+      "step": 22,
+      "tokens/total": 667632,
+      "tokens/train_per_sec_per_gpu": 36.14,
+      "tokens/trainable": 10086
+    },
+    {
+      "epoch": 0.08984375,
+      "grad_norm": 2.2993996143341064,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 0.05701298639178276,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.05867,
+      "step": 23,
+      "tokens/total": 697872,
+      "tokens/train_per_sec_per_gpu": 32.84,
+      "tokens/trainable": 10535
+    },
+    {
+      "epoch": 0.09375,
+      "grad_norm": 0.7493157386779785,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 0.022520896047353745,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.02278,
+      "step": 24,
+      "tokens/total": 728432,
+      "tokens/train_per_sec_per_gpu": 31.5,
+      "tokens/trainable": 10993
+    },
+    {
+      "epoch": 0.09765625,
+      "grad_norm": 0.7396641373634338,
+      "learning_rate": 9.6e-05,
+      "loss": 0.02083846554160118,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02106,
+      "step": 25,
+      "tokens/total": 756656,
+      "tokens/train_per_sec_per_gpu": 39.89,
+      "tokens/trainable": 11417
+    },
+    {
+      "epoch": 0.1015625,
+      "grad_norm": 1.161399483680725,
+      "learning_rate": 0.0001,
+      "loss": 0.028417319059371948,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.02882,
+      "step": 26,
+      "tokens/total": 787120,
+      "tokens/train_per_sec_per_gpu": 34.29,
+      "tokens/trainable": 11874
+    },
+    {
+      "epoch": 0.10546875,
+      "grad_norm": 0.4469190835952759,
+      "learning_rate": 9.99990636831587e-05,
+      "loss": 0.011035051196813583,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.0111,
+      "step": 27,
+      "tokens/total": 817504,
+      "tokens/train_per_sec_per_gpu": 31.64,
+      "tokens/trainable": 12316
+    },
+    {
+      "epoch": 0.109375,
+      "grad_norm": 1.2457350492477417,
+      "learning_rate": 9.999625477159879e-05,
+      "loss": 0.019333787262439728,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.01952,
+      "step": 28,
+      "tokens/total": 848128,
+      "tokens/train_per_sec_per_gpu": 37.77,
+      "tokens/trainable": 12800
+    },
+    {
+      "epoch": 0.11328125,
+      "grad_norm": 4.982752799987793,
+      "learning_rate": 9.999157338221051e-05,
+      "loss": 0.025677010416984558,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02601,
+      "step": 29,
+      "tokens/total": 878448,
+      "tokens/train_per_sec_per_gpu": 35.45,
+      "tokens/trainable": 13270
+    },
+    {
+      "epoch": 0.1171875,
+      "grad_norm": 0.3679017424583435,
+      "learning_rate": 9.998501970980562e-05,
+      "loss": 0.00844038650393486,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.00848,
+      "step": 30,
+      "tokens/total": 909008,
+      "tokens/train_per_sec_per_gpu": 35.3,
+      "tokens/trainable": 13752
+    },
+    {
+      "epoch": 0.12109375,
+      "grad_norm": 1.04750394821167,
+      "learning_rate": 9.997659402710915e-05,
+      "loss": 0.021330781280994415,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02156,
+      "step": 31,
+      "tokens/total": 939312,
+      "tokens/train_per_sec_per_gpu": 34.01,
+      "tokens/trainable": 14218
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 1.9312676191329956,
+      "learning_rate": 9.996629668474818e-05,
+      "loss": 0.010071363300085068,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01012,
+      "step": 32,
+      "tokens/total": 969264,
+      "tokens/train_per_sec_per_gpu": 34.43,
+      "tokens/trainable": 14655
+    },
+    {
+      "epoch": 0.12890625,
+      "grad_norm": 1.2505651712417603,
+      "learning_rate": 9.995412811123711e-05,
+      "loss": 0.05041354149580002,
+      "memory/device_reserved (GiB)": 37.82,
+      "memory/max_active (GiB)": 33.95,
+      "memory/max_allocated (GiB)": 33.95,
+      "ppl": 1.05171,
+      "step": 33,
+      "tokens/total": 999792,
+      "tokens/train_per_sec_per_gpu": 33.25,
+      "tokens/trainable": 15087
+    },
+    {
+      "epoch": 0.1328125,
+      "grad_norm": 1.8183362483978271,
+      "learning_rate": 9.994008881295999e-05,
+      "loss": 0.041947491466999054,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.04284,
+      "step": 34,
+      "tokens/total": 1029840,
+      "tokens/train_per_sec_per_gpu": 33.91,
+      "tokens/trainable": 15548
+    },
+    {
+      "epoch": 0.13671875,
+      "grad_norm": 0.8506646752357483,
+      "learning_rate": 9.992417937414932e-05,
+      "loss": 0.03074970841407776,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03123,
+      "step": 35,
+      "tokens/total": 1060416,
+      "tokens/train_per_sec_per_gpu": 33.72,
+      "tokens/trainable": 16037
+    },
+    {
+      "epoch": 0.140625,
+      "grad_norm": 0.8499165773391724,
+      "learning_rate": 9.99064004568618e-05,
+      "loss": 0.015379857271909714,
+      "memory/device_reserved (GiB)": 35.48,
+      "memory/max_active (GiB)": 33.72,
+      "memory/max_allocated (GiB)": 33.72,
+      "ppl": 1.0155,
+      "step": 36,
+      "tokens/total": 1090464,
+      "tokens/train_per_sec_per_gpu": 31.02,
+      "tokens/trainable": 16475
+    },
+    {
+      "epoch": 0.14453125,
+      "grad_norm": 0.824851393699646,
+      "learning_rate": 9.988675280095074e-05,
+      "loss": 0.03565571829676628,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.0363,
+      "step": 37,
+      "tokens/total": 1120944,
+      "tokens/train_per_sec_per_gpu": 34.29,
+      "tokens/trainable": 16919
+    },
+    {
+      "epoch": 0.1484375,
+      "grad_norm": 0.9401444792747498,
+      "learning_rate": 9.986523722403528e-05,
+      "loss": 0.03219553083181381,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.03272,
+      "step": 38,
+      "tokens/total": 1151360,
+      "tokens/train_per_sec_per_gpu": 28.92,
+      "tokens/trainable": 17333
+    },
+    {
+      "epoch": 0.15234375,
+      "grad_norm": 1.8253111839294434,
+      "learning_rate": 9.984185462146642e-05,
+      "loss": 0.04147548973560333,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.04235,
+      "step": 39,
+      "tokens/total": 1181728,
+      "tokens/train_per_sec_per_gpu": 36.1,
+      "tokens/trainable": 17806
+    },
+    {
+      "epoch": 0.15625,
+      "grad_norm": 2.6716220378875732,
+      "learning_rate": 9.98166059662897e-05,
+      "loss": 0.049233876168727875,
+      "memory/device_reserved (GiB)": 35.86,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.05047,
+      "step": 40,
+      "tokens/total": 1212064,
+      "tokens/train_per_sec_per_gpu": 35.55,
+      "tokens/trainable": 18290
+    },
+    {
+      "epoch": 0.16015625,
+      "grad_norm": 0.9621749520301819,
+      "learning_rate": 9.978949230920472e-05,
+      "loss": 0.03484003618359566,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.03545,
+      "step": 41,
+      "tokens/total": 1242592,
+      "tokens/train_per_sec_per_gpu": 31.95,
+      "tokens/trainable": 18747
+    },
+    {
+      "epoch": 0.1640625,
+      "grad_norm": 0.7557086944580078,
+      "learning_rate": 9.976051477852141e-05,
+      "loss": 0.025474129244685173,
+      "memory/device_reserved (GiB)": 35.88,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.0258,
+      "step": 42,
+      "tokens/total": 1272800,
+      "tokens/train_per_sec_per_gpu": 31.84,
+      "tokens/trainable": 19172
+    },
+    {
+      "epoch": 0.16796875,
+      "grad_norm": 0.20285196602344513,
+      "learning_rate": 9.972967458011312e-05,
+      "loss": 0.006379525177180767,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 34.04,
+      "memory/max_allocated (GiB)": 34.04,
+      "ppl": 1.0064,
+      "step": 43,
+      "tokens/total": 1303600,
+      "tokens/train_per_sec_per_gpu": 31.24,
+      "tokens/trainable": 19622
+    },
+    {
+      "epoch": 0.171875,
+      "grad_norm": 0.46316176652908325,
+      "learning_rate": 9.96969729973664e-05,
+      "loss": 0.01995321363210678,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02015,
+      "step": 44,
+      "tokens/total": 1333936,
+      "tokens/train_per_sec_per_gpu": 33.33,
+      "tokens/trainable": 20073
+    },
+    {
+      "epoch": 0.17578125,
+      "grad_norm": 2.533421516418457,
+      "learning_rate": 9.966241139112754e-05,
+      "loss": 0.027163950726389885,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.02754,
+      "step": 45,
+      "tokens/total": 1364288,
+      "tokens/train_per_sec_per_gpu": 31.41,
+      "tokens/trainable": 20497
+    },
+    {
+      "epoch": 0.1796875,
+      "grad_norm": 1.1473517417907715,
+      "learning_rate": 9.96259911996461e-05,
+      "loss": 0.023375950753688812,
+      "memory/device_reserved (GiB)": 36.03,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02365,
+      "step": 46,
+      "tokens/total": 1394368,
+      "tokens/train_per_sec_per_gpu": 34.23,
+      "tokens/trainable": 20940
+    },
+    {
+      "epoch": 0.18359375,
+      "grad_norm": 1.0396223068237305,
+      "learning_rate": 9.958771393851491e-05,
+      "loss": 0.011927608400583267,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.98,
+      "memory/max_allocated (GiB)": 33.98,
+      "ppl": 1.012,
+      "step": 47,
+      "tokens/total": 1425024,
+      "tokens/train_per_sec_per_gpu": 34.96,
+      "tokens/trainable": 21405
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 2.1290171146392822,
+      "learning_rate": 9.954758120060702e-05,
+      "loss": 0.00876394473016262,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.0088,
+      "step": 48,
+      "tokens/total": 1455392,
+      "tokens/train_per_sec_per_gpu": 34.94,
+      "tokens/trainable": 21828
+    },
+    {
+      "epoch": 0.19140625,
+      "grad_norm": 0.7430047988891602,
+      "learning_rate": 9.950559465600948e-05,
+      "loss": 0.010768749751150608,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01083,
+      "step": 49,
+      "tokens/total": 1485616,
+      "tokens/train_per_sec_per_gpu": 29.38,
+      "tokens/trainable": 22241
+    },
+    {
+      "epoch": 0.1953125,
+      "grad_norm": 0.5257824063301086,
+      "learning_rate": 9.946175605195379e-05,
+      "loss": 0.013969912193715572,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01407,
+      "step": 50,
+      "tokens/total": 1515984,
+      "tokens/train_per_sec_per_gpu": 36.31,
+      "tokens/trainable": 22688
+    },
+    {
+      "epoch": 0.19921875,
+      "grad_norm": 0.7611995339393616,
+      "learning_rate": 9.941606721274322e-05,
+      "loss": 0.007136075757443905,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00716,
+      "step": 51,
+      "tokens/total": 1546224,
+      "tokens/train_per_sec_per_gpu": 32.06,
+      "tokens/trainable": 23136
+    },
+    {
+      "epoch": 0.203125,
+      "grad_norm": 0.39372026920318604,
+      "learning_rate": 9.936853003967685e-05,
+      "loss": 0.008833611384034157,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00887,
+      "step": 52,
+      "tokens/total": 1576528,
+      "tokens/train_per_sec_per_gpu": 36.89,
+      "tokens/trainable": 23631
+    },
+    {
+      "epoch": 0.20703125,
+      "grad_norm": 0.6142369508743286,
+      "learning_rate": 9.93191465109705e-05,
+      "loss": 0.01554907951503992,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01567,
+      "step": 53,
+      "tokens/total": 1606768,
+      "tokens/train_per_sec_per_gpu": 34.11,
+      "tokens/trainable": 24096
+    },
+    {
+      "epoch": 0.2109375,
+      "grad_norm": 0.3934486210346222,
+      "learning_rate": 9.926791868167438e-05,
+      "loss": 0.012682640925049782,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.66,
+      "memory/max_allocated (GiB)": 33.66,
+      "ppl": 1.01276,
+      "step": 54,
+      "tokens/total": 1636640,
+      "tokens/train_per_sec_per_gpu": 37.8,
+      "tokens/trainable": 24569
+    },
+    {
+      "epoch": 0.21484375,
+      "grad_norm": 1.206893801689148,
+      "learning_rate": 9.921484868358753e-05,
+      "loss": 0.013577264733612537,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01367,
+      "step": 55,
+      "tokens/total": 1667136,
+      "tokens/train_per_sec_per_gpu": 33.9,
+      "tokens/trainable": 25067
+    },
+    {
+      "epoch": 0.21875,
+      "grad_norm": 0.1256696581840515,
+      "learning_rate": 9.915993872516924e-05,
+      "loss": 0.0021174070425331593,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00212,
+      "step": 56,
+      "tokens/total": 1697472,
+      "tokens/train_per_sec_per_gpu": 36.49,
+      "tokens/trainable": 25541
+    },
+    {
+      "epoch": 0.22265625,
+      "grad_norm": 0.3973124325275421,
+      "learning_rate": 9.9103191091447e-05,
+      "loss": 0.0054704658687114716,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00549,
+      "step": 57,
+      "tokens/total": 1727680,
+      "tokens/train_per_sec_per_gpu": 33.16,
+      "tokens/trainable": 25965
+    },
+    {
+      "epoch": 0.2265625,
+      "grad_norm": 1.6158578395843506,
+      "learning_rate": 9.904460814392147e-05,
+      "loss": 0.022541973739862442,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0228,
+      "step": 58,
+      "tokens/total": 1758192,
+      "tokens/train_per_sec_per_gpu": 33.85,
+      "tokens/trainable": 26409
+    },
+    {
+      "epoch": 0.23046875,
+      "grad_norm": 1.330061912536621,
+      "learning_rate": 9.898419232046825e-05,
+      "loss": 0.013835040852427483,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.01393,
+      "step": 59,
+      "tokens/total": 1788816,
+      "tokens/train_per_sec_per_gpu": 33.99,
+      "tokens/trainable": 26883
+    },
+    {
+      "epoch": 0.234375,
+      "grad_norm": 0.7960736751556396,
+      "learning_rate": 9.892194613523633e-05,
+      "loss": 0.011244643479585648,
+      "memory/device_reserved (GiB)": 36.15,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01131,
+      "step": 60,
+      "tokens/total": 1819056,
+      "tokens/train_per_sec_per_gpu": 33.66,
+      "tokens/trainable": 27323
+    },
+    {
+      "epoch": 0.23828125,
+      "grad_norm": 0.6835820078849792,
+      "learning_rate": 9.885787217854357e-05,
+      "loss": 0.00427593570202589,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00429,
+      "step": 61,
+      "tokens/total": 1849664,
+      "tokens/train_per_sec_per_gpu": 39.15,
+      "tokens/trainable": 27830
+    },
+    {
+      "epoch": 0.2421875,
+      "grad_norm": 2.0484817028045654,
+      "learning_rate": 9.879197311676887e-05,
+      "loss": 0.034456316381692886,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 34.02,
+      "memory/max_allocated (GiB)": 34.02,
+      "ppl": 1.03506,
+      "step": 62,
+      "tokens/total": 1880272,
+      "tokens/train_per_sec_per_gpu": 32.25,
+      "tokens/trainable": 28281
+    },
+    {
+      "epoch": 0.24609375,
+      "grad_norm": 1.2015022039413452,
+      "learning_rate": 9.872425169224113e-05,
+      "loss": 0.021808233112096786,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.02205,
+      "step": 63,
+      "tokens/total": 1910752,
+      "tokens/train_per_sec_per_gpu": 32.14,
+      "tokens/trainable": 28742
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3692668378353119,
+      "learning_rate": 9.865471072312528e-05,
+      "loss": 0.010146009735763073,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.0102,
+      "step": 64,
+      "tokens/total": 1940848,
+      "tokens/train_per_sec_per_gpu": 33.53,
+      "tokens/trainable": 29183
+    },
+    {
+      "epoch": 0.25390625,
+      "grad_norm": 0.5080033540725708,
+      "learning_rate": 9.858335310330492e-05,
+      "loss": 0.015261693857610226,
+      "memory/device_reserved (GiB)": 36.16,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.01538,
+      "step": 65,
+      "tokens/total": 1971344,
+      "tokens/train_per_sec_per_gpu": 31.43,
+      "tokens/trainable": 29648
+    },
+    {
+      "epoch": 0.2578125,
+      "grad_norm": 9.397146224975586,
+      "learning_rate": 9.851018180226185e-05,
+      "loss": 0.03658987954258919,
+      "memory/device_reserved (GiB)": 35.06,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.03727,
+      "step": 66,
+      "tokens/total": 2001712,
+      "tokens/train_per_sec_per_gpu": 33.39,
+      "tokens/trainable": 30075
+    },
+    {
+      "epoch": 0.26171875,
+      "grad_norm": 0.8944637775421143,
+      "learning_rate": 9.843519986495259e-05,
+      "loss": 0.010249090380966663,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.0103,
+      "step": 67,
+      "tokens/total": 2029936,
+      "tokens/train_per_sec_per_gpu": 37.46,
+      "tokens/trainable": 30546
+    },
+    {
+      "epoch": 0.265625,
+      "grad_norm": 0.2694835066795349,
+      "learning_rate": 9.835841041168162e-05,
+      "loss": 0.0061509087681770325,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00617,
+      "step": 68,
+      "tokens/total": 2060384,
+      "tokens/train_per_sec_per_gpu": 35.03,
+      "tokens/trainable": 31025
+    },
+    {
+      "epoch": 0.26953125,
+      "grad_norm": 1.0974539518356323,
+      "learning_rate": 9.82798166379715e-05,
+      "loss": 0.005264772102236748,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00528,
+      "step": 69,
+      "tokens/total": 2090688,
+      "tokens/train_per_sec_per_gpu": 33.74,
+      "tokens/trainable": 31505
+    },
+    {
+      "epoch": 0.2734375,
+      "grad_norm": 0.6211183667182922,
+      "learning_rate": 9.819942181443002e-05,
+      "loss": 0.008906489238142967,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.00895,
+      "step": 70,
+      "tokens/total": 2120848,
+      "tokens/train_per_sec_per_gpu": 32.2,
+      "tokens/trainable": 31956
+    },
+    {
+      "epoch": 0.27734375,
+      "grad_norm": 0.13163211941719055,
+      "learning_rate": 9.811722928661392e-05,
+      "loss": 0.0017089881002902985,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.35,
+      "memory/max_allocated (GiB)": 33.35,
+      "ppl": 1.00171,
+      "step": 71,
+      "tokens/total": 2149120,
+      "tokens/train_per_sec_per_gpu": 33.21,
+      "tokens/trainable": 32374
+    },
+    {
+      "epoch": 0.28125,
+      "grad_norm": 1.1125941276550293,
+      "learning_rate": 9.803324247488975e-05,
+      "loss": 0.026288464665412903,
+      "memory/device_reserved (GiB)": 35.64,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02664,
+      "step": 72,
+      "tokens/total": 2179472,
+      "tokens/train_per_sec_per_gpu": 34.41,
+      "tokens/trainable": 32833
+    },
+    {
+      "epoch": 0.28515625,
+      "grad_norm": 0.4967571198940277,
+      "learning_rate": 9.794746487429161e-05,
+      "loss": 0.007215567864477634,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00724,
+      "step": 73,
+      "tokens/total": 2209984,
+      "tokens/train_per_sec_per_gpu": 35.63,
+      "tokens/trainable": 33304
+    },
+    {
+      "epoch": 0.2890625,
+      "grad_norm": 1.2762513160705566,
+      "learning_rate": 9.785990005437554e-05,
+      "loss": 0.007767696864902973,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.0078,
+      "step": 74,
+      "tokens/total": 2240512,
+      "tokens/train_per_sec_per_gpu": 32.3,
+      "tokens/trainable": 33771
+    },
+    {
+      "epoch": 0.29296875,
+      "grad_norm": 0.687618613243103,
+      "learning_rate": 9.777055165907117e-05,
+      "loss": 0.010631063021719456,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01069,
+      "step": 75,
+      "tokens/total": 2270608,
+      "tokens/train_per_sec_per_gpu": 34.43,
+      "tokens/trainable": 34186
+    },
+    {
+      "epoch": 0.296875,
+      "grad_norm": 0.47425633668899536,
+      "learning_rate": 9.767942340652993e-05,
+      "loss": 0.007360030896961689,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00739,
+      "step": 76,
+      "tokens/total": 2300864,
+      "tokens/train_per_sec_per_gpu": 34.79,
+      "tokens/trainable": 34667
+    },
+    {
+      "epoch": 0.30078125,
+      "grad_norm": 0.8886299133300781,
+      "learning_rate": 9.758651908897035e-05,
+      "loss": 0.07993505895137787,
+      "memory/device_reserved (GiB)": 35.96,
+      "memory/max_active (GiB)": 33.76,
+      "memory/max_allocated (GiB)": 33.76,
+      "ppl": 1.08322,
+      "step": 77,
+      "tokens/total": 2330960,
+      "tokens/train_per_sec_per_gpu": 31.68,
+      "tokens/trainable": 35078
+    },
+    {
+      "epoch": 0.3046875,
+      "grad_norm": 0.7701996564865112,
+      "learning_rate": 9.749184257252033e-05,
+      "loss": 0.006262673996388912,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.00628,
+      "step": 78,
+      "tokens/total": 2361584,
+      "tokens/train_per_sec_per_gpu": 34.76,
+      "tokens/trainable": 35550
+    },
+    {
+      "epoch": 0.30859375,
+      "grad_norm": 0.4809786379337311,
+      "learning_rate": 9.739539779705614e-05,
+      "loss": 0.005311621818691492,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00533,
+      "step": 79,
+      "tokens/total": 2391936,
+      "tokens/train_per_sec_per_gpu": 29.42,
+      "tokens/trainable": 35977
+    },
+    {
+      "epoch": 0.3125,
+      "grad_norm": 0.7056035995483398,
+      "learning_rate": 9.729718877603861e-05,
+      "loss": 0.019523054361343384,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01971,
+      "step": 80,
+      "tokens/total": 2422432,
+      "tokens/train_per_sec_per_gpu": 38.13,
+      "tokens/trainable": 36462
+    },
+    {
+      "epoch": 0.31640625,
+      "grad_norm": 1.1266603469848633,
+      "learning_rate": 9.719721959634592e-05,
+      "loss": 0.016325978562235832,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01646,
+      "step": 81,
+      "tokens/total": 2452752,
+      "tokens/train_per_sec_per_gpu": 33.21,
+      "tokens/trainable": 36919
+    },
+    {
+      "epoch": 0.3203125,
+      "grad_norm": 1.0165926218032837,
+      "learning_rate": 9.709549441810375e-05,
+      "loss": 0.01072955783456564,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.44,
+      "memory/max_allocated (GiB)": 33.44,
+      "ppl": 1.01079,
+      "step": 82,
+      "tokens/total": 2481216,
+      "tokens/train_per_sec_per_gpu": 33.42,
+      "tokens/trainable": 37379
+    },
+    {
+      "epoch": 0.32421875,
+      "grad_norm": 0.9835495948791504,
+      "learning_rate": 9.699201747451195e-05,
+      "loss": 0.01992407627403736,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.02012,
+      "step": 83,
+      "tokens/total": 2511488,
+      "tokens/train_per_sec_per_gpu": 33.04,
+      "tokens/trainable": 37832
+    },
+    {
+      "epoch": 0.328125,
+      "grad_norm": 1.3131390810012817,
+      "learning_rate": 9.688679307166854e-05,
+      "loss": 0.02011384442448616,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.02032,
+      "step": 84,
+      "tokens/total": 2541984,
+      "tokens/train_per_sec_per_gpu": 33.9,
+      "tokens/trainable": 38303
+    },
+    {
+      "epoch": 0.33203125,
+      "grad_norm": 2.3927090167999268,
+      "learning_rate": 9.677982558839042e-05,
+      "loss": 0.01429318729788065,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.69,
+      "memory/max_allocated (GiB)": 33.69,
+      "ppl": 1.0144,
+      "step": 85,
+      "tokens/total": 2572224,
+      "tokens/train_per_sec_per_gpu": 33.59,
+      "tokens/trainable": 38758
+    },
+    {
+      "epoch": 0.3359375,
+      "grad_norm": 0.7415490746498108,
+      "learning_rate": 9.66711194760312e-05,
+      "loss": 0.019946210086345673,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.02015,
+      "step": 86,
+      "tokens/total": 2602704,
+      "tokens/train_per_sec_per_gpu": 30.6,
+      "tokens/trainable": 39179
+    },
+    {
+      "epoch": 0.33984375,
+      "grad_norm": 0.8350754976272583,
+      "learning_rate": 9.656067925829593e-05,
+      "loss": 0.019459044560790062,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01965,
+      "step": 87,
+      "tokens/total": 2633072,
+      "tokens/train_per_sec_per_gpu": 35.28,
+      "tokens/trainable": 39679
+    },
+    {
+      "epoch": 0.34375,
+      "grad_norm": 0.9457952976226807,
+      "learning_rate": 9.644850953105288e-05,
+      "loss": 0.02666199952363968,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.02702,
+      "step": 88,
+      "tokens/total": 2663232,
+      "tokens/train_per_sec_per_gpu": 32.67,
+      "tokens/trainable": 40118
+    },
+    {
+      "epoch": 0.34765625,
+      "grad_norm": 0.4730850160121918,
+      "learning_rate": 9.633461496214225e-05,
+      "loss": 0.020438212901353836,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.02065,
+      "step": 89,
+      "tokens/total": 2693696,
+      "tokens/train_per_sec_per_gpu": 32.82,
+      "tokens/trainable": 40571
+    },
+    {
+      "epoch": 0.3515625,
+      "grad_norm": 0.8041865229606628,
+      "learning_rate": 9.621900029118195e-05,
+      "loss": 0.021355140954256058,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02158,
+      "step": 90,
+      "tokens/total": 2723984,
+      "tokens/train_per_sec_per_gpu": 30.81,
+      "tokens/trainable": 40997
+    },
+    {
+      "epoch": 0.35546875,
+      "grad_norm": 0.15726213157176971,
+      "learning_rate": 9.610167032937036e-05,
+      "loss": 0.0043283673003315926,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00434,
+      "step": 91,
+      "tokens/total": 2754240,
+      "tokens/train_per_sec_per_gpu": 36.39,
+      "tokens/trainable": 41462
+    },
+    {
+      "epoch": 0.359375,
+      "grad_norm": 1.0282626152038574,
+      "learning_rate": 9.598262995928611e-05,
+      "loss": 0.014904867857694626,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01502,
+      "step": 92,
+      "tokens/total": 2784672,
+      "tokens/train_per_sec_per_gpu": 36.84,
+      "tokens/trainable": 41927
+    },
+    {
+      "epoch": 0.36328125,
+      "grad_norm": 0.2766452133655548,
+      "learning_rate": 9.586188413468492e-05,
+      "loss": 0.0071023497730493546,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00713,
+      "step": 93,
+      "tokens/total": 2815120,
+      "tokens/train_per_sec_per_gpu": 34.85,
+      "tokens/trainable": 42397
+    },
+    {
+      "epoch": 0.3671875,
+      "grad_norm": 0.9832137823104858,
+      "learning_rate": 9.57394378802934e-05,
+      "loss": 0.011417705565690994,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01148,
+      "step": 94,
+      "tokens/total": 2845600,
+      "tokens/train_per_sec_per_gpu": 36.54,
+      "tokens/trainable": 42883
+    },
+    {
+      "epoch": 0.37109375,
+      "grad_norm": 0.4514453709125519,
+      "learning_rate": 9.56152962916e-05,
+      "loss": 0.007363555021584034,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00739,
+      "step": 95,
+      "tokens/total": 2875952,
+      "tokens/train_per_sec_per_gpu": 38.7,
+      "tokens/trainable": 43381
+    },
+    {
+      "epoch": 0.375,
+      "grad_norm": 0.555672287940979,
+      "learning_rate": 9.548946453464296e-05,
+      "loss": 0.00891757383942604,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00896,
+      "step": 96,
+      "tokens/total": 2906288,
+      "tokens/train_per_sec_per_gpu": 33.01,
+      "tokens/trainable": 43824
+    },
+    {
+      "epoch": 0.37890625,
+      "grad_norm": 0.37105968594551086,
+      "learning_rate": 9.53619478457953e-05,
+      "loss": 0.007098545785993338,
+      "memory/device_reserved (GiB)": 35.98,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00712,
+      "step": 97,
+      "tokens/total": 2936752,
+      "tokens/train_per_sec_per_gpu": 31.48,
+      "tokens/trainable": 44255
+    },
+    {
+      "epoch": 0.3828125,
+      "grad_norm": 1.0755594968795776,
+      "learning_rate": 9.523275153154695e-05,
+      "loss": 0.028615131974220276,
+      "memory/device_reserved (GiB)": 35.19,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.02903,
+      "step": 98,
+      "tokens/total": 2967072,
+      "tokens/train_per_sec_per_gpu": 37.66,
+      "tokens/trainable": 44739
+    },
+    {
+      "epoch": 0.38671875,
+      "grad_norm": 0.9296021461486816,
+      "learning_rate": 9.51018809682839e-05,
+      "loss": 0.01639867201447487,
+      "memory/device_reserved (GiB)": 35.76,
+      "memory/max_active (GiB)": 34.03,
+      "memory/max_allocated (GiB)": 34.03,
+      "ppl": 1.01653,
+      "step": 99,
+      "tokens/total": 2997664,
+      "tokens/train_per_sec_per_gpu": 36.17,
+      "tokens/trainable": 45202
+    },
+    {
+      "epoch": 0.390625,
+      "grad_norm": 0.6913011074066162,
+      "learning_rate": 9.49693416020645e-05,
+      "loss": 0.015926485881209373,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01605,
+      "step": 100,
+      "tokens/total": 3028032,
+      "tokens/train_per_sec_per_gpu": 33.06,
+      "tokens/trainable": 45646
+    },
+    {
+      "epoch": 0.39453125,
+      "grad_norm": 0.3258785605430603,
+      "learning_rate": 9.483513894839276e-05,
+      "loss": 0.008693980053067207,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00873,
+      "step": 101,
+      "tokens/total": 3058272,
+      "tokens/train_per_sec_per_gpu": 33.06,
+      "tokens/trainable": 46089
+    },
+    {
+      "epoch": 0.3984375,
+      "grad_norm": 1.0787935256958008,
+      "learning_rate": 9.469927859198888e-05,
+      "loss": 0.01954980194568634,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.01974,
+      "step": 102,
+      "tokens/total": 3088848,
+      "tokens/train_per_sec_per_gpu": 32.59,
+      "tokens/trainable": 46547
+    },
+    {
+      "epoch": 0.40234375,
+      "grad_norm": 0.9061365127563477,
+      "learning_rate": 9.456176618655689e-05,
+      "loss": 0.03529620170593262,
+      "memory/device_reserved (GiB)": 35.99,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.03593,
+      "step": 103,
+      "tokens/total": 3119424,
+      "tokens/train_per_sec_per_gpu": 31.97,
+      "tokens/trainable": 47014
+    },
+    {
+      "epoch": 0.40625,
+      "grad_norm": 2.671016216278076,
+      "learning_rate": 9.442260745454927e-05,
+      "loss": 0.02147374488413334,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.02171,
+      "step": 104,
+      "tokens/total": 3149696,
+      "tokens/train_per_sec_per_gpu": 37.68,
+      "tokens/trainable": 47503
+    },
+    {
+      "epoch": 0.41015625,
+      "grad_norm": 0.26923975348472595,
+      "learning_rate": 9.428180818692884e-05,
+      "loss": 0.007999668829143047,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00803,
+      "step": 105,
+      "tokens/total": 3180064,
+      "tokens/train_per_sec_per_gpu": 36.28,
+      "tokens/trainable": 47996
+    },
+    {
+      "epoch": 0.4140625,
+      "grad_norm": 0.7145718932151794,
+      "learning_rate": 9.413937424292791e-05,
+      "loss": 0.030988795682787895,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.03147,
+      "step": 106,
+      "tokens/total": 3208320,
+      "tokens/train_per_sec_per_gpu": 34.04,
+      "tokens/trainable": 48447
+    },
+    {
+      "epoch": 0.41796875,
+      "grad_norm": 0.20594321191310883,
+      "learning_rate": 9.399531154980424e-05,
+      "loss": 0.007135331630706787,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.00716,
+      "step": 107,
+      "tokens/total": 3238416,
+      "tokens/train_per_sec_per_gpu": 30.69,
+      "tokens/trainable": 48889
+    },
+    {
+      "epoch": 0.421875,
+      "grad_norm": 0.2812410891056061,
+      "learning_rate": 9.384962610259455e-05,
+      "loss": 0.010571298189461231,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.01063,
+      "step": 108,
+      "tokens/total": 3268784,
+      "tokens/train_per_sec_per_gpu": 37.9,
+      "tokens/trainable": 49360
+    },
+    {
+      "epoch": 0.42578125,
+      "grad_norm": 0.3485014736652374,
+      "learning_rate": 9.370232396386494e-05,
+      "loss": 0.016001982614398003,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.01613,
+      "step": 109,
+      "tokens/total": 3298912,
+      "tokens/train_per_sec_per_gpu": 36.73,
+      "tokens/trainable": 49827
+    },
+    {
+      "epoch": 0.4296875,
+      "grad_norm": 1.474901795387268,
+      "learning_rate": 9.355341126345868e-05,
+      "loss": 0.024302365258336067,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.0246,
+      "step": 110,
+      "tokens/total": 3329232,
+      "tokens/train_per_sec_per_gpu": 34.23,
+      "tokens/trainable": 50267
+    },
+    {
+      "epoch": 0.43359375,
+      "grad_norm": 0.5579241514205933,
+      "learning_rate": 9.340289419824107e-05,
+      "loss": 0.017602810636162758,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.01776,
+      "step": 111,
+      "tokens/total": 3359440,
+      "tokens/train_per_sec_per_gpu": 33.34,
+      "tokens/trainable": 50711
+    },
+    {
+      "epoch": 0.4375,
+      "grad_norm": 0.5616764426231384,
+      "learning_rate": 9.325077903184159e-05,
+      "loss": 0.015218488872051239,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01533,
+      "step": 112,
+      "tokens/total": 3389472,
+      "tokens/train_per_sec_per_gpu": 31.58,
+      "tokens/trainable": 51157
+    },
+    {
+      "epoch": 0.44140625,
+      "grad_norm": 0.38610321283340454,
+      "learning_rate": 9.30970720943932e-05,
+      "loss": 0.008289901539683342,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.00832,
+      "step": 113,
+      "tokens/total": 3419920,
+      "tokens/train_per_sec_per_gpu": 32.67,
+      "tokens/trainable": 51613
+    },
+    {
+      "epoch": 0.4453125,
+      "grad_norm": 0.3681856095790863,
+      "learning_rate": 9.2941779782269e-05,
+      "loss": 0.022977739572525024,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.02324,
+      "step": 114,
+      "tokens/total": 3450176,
+      "tokens/train_per_sec_per_gpu": 29.07,
+      "tokens/trainable": 52017
+    },
+    {
+      "epoch": 0.44921875,
+      "grad_norm": 0.2453298270702362,
+      "learning_rate": 9.278490855781596e-05,
+      "loss": 0.004594211932271719,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.0046,
+      "step": 115,
+      "tokens/total": 3480544,
+      "tokens/train_per_sec_per_gpu": 36.22,
+      "tokens/trainable": 52506
+    },
+    {
+      "epoch": 0.453125,
+      "grad_norm": 0.37761810421943665,
+      "learning_rate": 9.262646494908604e-05,
+      "loss": 0.011891474947333336,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.01196,
+      "step": 116,
+      "tokens/total": 3510848,
+      "tokens/train_per_sec_per_gpu": 38.57,
+      "tokens/trainable": 53008
+    },
+    {
+      "epoch": 0.45703125,
+      "grad_norm": 1.953158974647522,
+      "learning_rate": 9.246645554956457e-05,
+      "loss": 0.006801537703722715,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.96,
+      "memory/max_allocated (GiB)": 33.96,
+      "ppl": 1.00682,
+      "step": 117,
+      "tokens/total": 3541296,
+      "tokens/train_per_sec_per_gpu": 35.38,
+      "tokens/trainable": 53467
+    },
+    {
+      "epoch": 0.4609375,
+      "grad_norm": 0.45078471302986145,
+      "learning_rate": 9.230488701789578e-05,
+      "loss": 0.017124727368354797,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.01727,
+      "step": 118,
+      "tokens/total": 3571936,
+      "tokens/train_per_sec_per_gpu": 32.83,
+      "tokens/trainable": 53923
+    },
+    {
+      "epoch": 0.46484375,
+      "grad_norm": 0.6464744806289673,
+      "learning_rate": 9.214176607760577e-05,
+      "loss": 0.012542763724923134,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.01262,
+      "step": 119,
+      "tokens/total": 3602000,
+      "tokens/train_per_sec_per_gpu": 35.6,
+      "tokens/trainable": 54363
+    },
+    {
+      "epoch": 0.46875,
+      "grad_norm": 0.3361028730869293,
+      "learning_rate": 9.197709951682268e-05,
+      "loss": 0.004235772415995598,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00424,
+      "step": 120,
+      "tokens/total": 3632288,
+      "tokens/train_per_sec_per_gpu": 35.26,
+      "tokens/trainable": 54839
+    },
+    {
+      "epoch": 0.47265625,
+      "grad_norm": 0.4455215036869049,
+      "learning_rate": 9.181089418799428e-05,
+      "loss": 0.004468773491680622,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00448,
+      "step": 121,
+      "tokens/total": 3662608,
+      "tokens/train_per_sec_per_gpu": 32.62,
+      "tokens/trainable": 55267
+    },
+    {
+      "epoch": 0.4765625,
+      "grad_norm": 0.3547821044921875,
+      "learning_rate": 9.164315700760271e-05,
+      "loss": 0.010471776127815247,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.01053,
+      "step": 122,
+      "tokens/total": 3692864,
+      "tokens/train_per_sec_per_gpu": 30.54,
+      "tokens/trainable": 55700
+    },
+    {
+      "epoch": 0.48046875,
+      "grad_norm": 0.7816572189331055,
+      "learning_rate": 9.147389495587671e-05,
+      "loss": 0.01556549035012722,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.73,
+      "memory/max_allocated (GiB)": 33.73,
+      "ppl": 1.01569,
+      "step": 123,
+      "tokens/total": 3722864,
+      "tokens/train_per_sec_per_gpu": 33.41,
+      "tokens/trainable": 56144
+    },
+    {
+      "epoch": 0.484375,
+      "grad_norm": 0.3188917338848114,
+      "learning_rate": 9.130311507650116e-05,
+      "loss": 0.0064091756939888,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.77,
+      "memory/max_allocated (GiB)": 33.77,
+      "ppl": 1.00643,
+      "step": 124,
+      "tokens/total": 3753056,
+      "tokens/train_per_sec_per_gpu": 33.13,
+      "tokens/trainable": 56589
+    },
+    {
+      "epoch": 0.48828125,
+      "grad_norm": 0.7098766565322876,
+      "learning_rate": 9.113082447632394e-05,
+      "loss": 0.026026558130979538,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.02637,
+      "step": 125,
+      "tokens/total": 3783312,
+      "tokens/train_per_sec_per_gpu": 36.6,
+      "tokens/trainable": 57081
+    },
+    {
+      "epoch": 0.4921875,
+      "grad_norm": 0.4195394217967987,
+      "learning_rate": 9.09570303250602e-05,
+      "loss": 0.01531171053647995,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01543,
+      "step": 126,
+      "tokens/total": 3813712,
+      "tokens/train_per_sec_per_gpu": 32.83,
+      "tokens/trainable": 57527
+    },
+    {
+      "epoch": 0.49609375,
+      "grad_norm": 0.4466935992240906,
+      "learning_rate": 9.078173985499394e-05,
+      "loss": 0.014355337247252464,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.01446,
+      "step": 127,
+      "tokens/total": 3844336,
+      "tokens/train_per_sec_per_gpu": 34.84,
+      "tokens/trainable": 58005
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.19709816575050354,
+      "learning_rate": 9.060496036067713e-05,
+      "loss": 0.006424291059374809,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.00644,
+      "step": 128,
+      "tokens/total": 3872688,
+      "tokens/train_per_sec_per_gpu": 35.24,
+      "tokens/trainable": 58454
+    },
+    {
+      "epoch": 0.50390625,
+      "grad_norm": 0.18546900153160095,
+      "learning_rate": 9.042669919862615e-05,
+      "loss": 0.003899412928149104,
+      "memory/device_reserved (GiB)": 36.0,
+      "memory/max_active (GiB)": 33.64,
+      "memory/max_allocated (GiB)": 33.64,
+      "ppl": 1.00391,
+      "step": 129,
+      "tokens/total": 3902640,
+      "tokens/train_per_sec_per_gpu": 29.88,
+      "tokens/trainable": 58896
+    },
+    {
+      "epoch": 0.5078125,
+      "grad_norm": 0.2095804512500763,
+      "learning_rate": 9.024696378701557e-05,
+      "loss": 0.005734554957598448,
+      "memory/device_reserved (GiB)": 34.68,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00575,
+      "step": 130,
+      "tokens/total": 3933104,
+      "tokens/train_per_sec_per_gpu": 35.88,
+      "tokens/trainable": 59356
+    },
+    {
+      "epoch": 0.51171875,
+      "grad_norm": 0.3548737168312073,
+      "learning_rate": 9.006576160536948e-05,
+      "loss": 0.010866682976484299,
+      "memory/device_reserved (GiB)": 34.92,
+      "memory/max_active (GiB)": 33.78,
+      "memory/max_allocated (GiB)": 33.78,
+      "ppl": 1.01093,
+      "step": 131,
+      "tokens/total": 3963184,
+      "tokens/train_per_sec_per_gpu": 32.18,
+      "tokens/trainable": 59792
+    },
+    {
+      "epoch": 0.515625,
+      "grad_norm": 0.24394433200359344,
+      "learning_rate": 8.988310019425035e-05,
+      "loss": 0.004735369700938463,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00475,
+      "step": 132,
+      "tokens/total": 3993648,
+      "tokens/train_per_sec_per_gpu": 35.71,
+      "tokens/trainable": 60266
+    },
+    {
+      "epoch": 0.51953125,
+      "grad_norm": 0.3678644597530365,
+      "learning_rate": 8.969898715494506e-05,
+      "loss": 0.007982447743415833,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00801,
+      "step": 133,
+      "tokens/total": 4024000,
+      "tokens/train_per_sec_per_gpu": 35.45,
+      "tokens/trainable": 60759
+    },
+    {
+      "epoch": 0.5234375,
+      "grad_norm": 0.6605305075645447,
+      "learning_rate": 8.951343014914869e-05,
+      "loss": 0.02931230701506138,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.39,
+      "memory/max_allocated (GiB)": 33.39,
+      "ppl": 1.02975,
+      "step": 134,
+      "tokens/total": 4052480,
+      "tokens/train_per_sec_per_gpu": 31.78,
+      "tokens/trainable": 61204
+    },
+    {
+      "epoch": 0.52734375,
+      "grad_norm": 0.1193099245429039,
+      "learning_rate": 8.932643689864568e-05,
+      "loss": 0.0025630262680351734,
+      "memory/device_reserved (GiB)": 35.78,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00257,
+      "step": 135,
+      "tokens/total": 4082864,
+      "tokens/train_per_sec_per_gpu": 35.97,
+      "tokens/trainable": 61684
+    },
+    {
+      "epoch": 0.53125,
+      "grad_norm": 0.12960873544216156,
+      "learning_rate": 8.913801518498845e-05,
+      "loss": 0.0019581567030400038,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.7,
+      "memory/max_allocated (GiB)": 33.7,
+      "ppl": 1.00196,
+      "step": 136,
+      "tokens/total": 4113040,
+      "tokens/train_per_sec_per_gpu": 38.25,
+      "tokens/trainable": 62159
+    },
+    {
+      "epoch": 0.53515625,
+      "grad_norm": 0.34460127353668213,
+      "learning_rate": 8.894817284917364e-05,
+      "loss": 0.01051283534616232,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.01057,
+      "step": 137,
+      "tokens/total": 4143344,
+      "tokens/train_per_sec_per_gpu": 35.31,
+      "tokens/trainable": 62649
+    },
+    {
+      "epoch": 0.5390625,
+      "grad_norm": 0.20421279966831207,
+      "learning_rate": 8.875691779131569e-05,
+      "loss": 0.0035965177230536938,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.0036,
+      "step": 138,
+      "tokens/total": 4173648,
+      "tokens/train_per_sec_per_gpu": 36.67,
+      "tokens/trainable": 63100
+    },
+    {
+      "epoch": 0.54296875,
+      "grad_norm": 0.30597925186157227,
+      "learning_rate": 8.856425797031829e-05,
+      "loss": 0.008190772496163845,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.92,
+      "memory/max_allocated (GiB)": 33.92,
+      "ppl": 1.00822,
+      "step": 139,
+      "tokens/total": 4204080,
+      "tokens/train_per_sec_per_gpu": 37.61,
+      "tokens/trainable": 63588
+    },
+    {
+      "epoch": 0.546875,
+      "grad_norm": 0.5688289999961853,
+      "learning_rate": 8.837020140354295e-05,
+      "loss": 0.022904174402356148,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.42,
+      "memory/max_allocated (GiB)": 33.42,
+      "ppl": 1.02317,
+      "step": 140,
+      "tokens/total": 4232400,
+      "tokens/train_per_sec_per_gpu": 33.09,
+      "tokens/trainable": 64005
+    },
+    {
+      "epoch": 0.55078125,
+      "grad_norm": 0.5879082083702087,
+      "learning_rate": 8.817475616647554e-05,
+      "loss": 0.013611489906907082,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.0137,
+      "step": 141,
+      "tokens/total": 4262624,
+      "tokens/train_per_sec_per_gpu": 35.14,
+      "tokens/trainable": 64462
+    },
+    {
+      "epoch": 0.5546875,
+      "grad_norm": 0.31034204363822937,
+      "learning_rate": 8.797793039239017e-05,
+      "loss": 0.008315861225128174,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.94,
+      "memory/max_allocated (GiB)": 33.94,
+      "ppl": 1.00835,
+      "step": 142,
+      "tokens/total": 4293168,
+      "tokens/train_per_sec_per_gpu": 33.95,
+      "tokens/trainable": 64900
+    },
+    {
+      "epoch": 0.55859375,
+      "grad_norm": 0.7010869979858398,
+      "learning_rate": 8.777973227201069e-05,
+      "loss": 0.022431056946516037,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.02268,
+      "step": 143,
+      "tokens/total": 4323248,
+      "tokens/train_per_sec_per_gpu": 37.57,
+      "tokens/trainable": 65379
+    },
+    {
+      "epoch": 0.5625,
+      "grad_norm": 0.2407914400100708,
+      "learning_rate": 8.758017005316988e-05,
+      "loss": 0.006207741796970367,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00623,
+      "step": 144,
+      "tokens/total": 4353344,
+      "tokens/train_per_sec_per_gpu": 36.72,
+      "tokens/trainable": 65845
+    },
+    {
+      "epoch": 0.56640625,
+      "grad_norm": 0.21913351118564606,
+      "learning_rate": 8.737925204046629e-05,
+      "loss": 0.009031139314174652,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00907,
+      "step": 145,
+      "tokens/total": 4383584,
+      "tokens/train_per_sec_per_gpu": 32.92,
+      "tokens/trainable": 66285
+    },
+    {
+      "epoch": 0.5703125,
+      "grad_norm": 0.39635229110717773,
+      "learning_rate": 8.717698659491851e-05,
+      "loss": 0.016251137480139732,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.74,
+      "memory/max_allocated (GiB)": 33.74,
+      "ppl": 1.01638,
+      "step": 146,
+      "tokens/total": 4413856,
+      "tokens/train_per_sec_per_gpu": 31.48,
+      "tokens/trainable": 66736
+    },
+    {
+      "epoch": 0.57421875,
+      "grad_norm": 0.5958674550056458,
+      "learning_rate": 8.697338213361735e-05,
+      "loss": 0.018637506291270256,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.01881,
+      "step": 147,
+      "tokens/total": 4444112,
+      "tokens/train_per_sec_per_gpu": 35.28,
+      "tokens/trainable": 67150
+    },
+    {
+      "epoch": 0.578125,
+      "grad_norm": 0.19332216680049896,
+      "learning_rate": 8.676844712937552e-05,
+      "loss": 0.006893862038850784,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00692,
+      "step": 148,
+      "tokens/total": 4474464,
+      "tokens/train_per_sec_per_gpu": 35.09,
+      "tokens/trainable": 67633
+    },
+    {
+      "epoch": 0.58203125,
+      "grad_norm": 0.36499589681625366,
+      "learning_rate": 8.656219011037509e-05,
+      "loss": 0.010607698000967503,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.01066,
+      "step": 149,
+      "tokens/total": 4504880,
+      "tokens/train_per_sec_per_gpu": 33.87,
+      "tokens/trainable": 68077
+    },
+    {
+      "epoch": 0.5859375,
+      "grad_norm": 0.10280518978834152,
+      "learning_rate": 8.63546196598125e-05,
+      "loss": 0.002519096015021205,
+      "memory/device_reserved (GiB)": 35.79,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00252,
+      "step": 150,
+      "tokens/total": 4534976,
+      "tokens/train_per_sec_per_gpu": 33.26,
+      "tokens/trainable": 68545
+    },
+    {
+      "epoch": 0.58984375,
+      "grad_norm": 0.24021312594413757,
+      "learning_rate": 8.614574441554145e-05,
+      "loss": 0.006152496673166752,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00617,
+      "step": 151,
+      "tokens/total": 4565088,
+      "tokens/train_per_sec_per_gpu": 30.26,
+      "tokens/trainable": 69011
+    },
+    {
+      "epoch": 0.59375,
+      "grad_norm": 0.5905681252479553,
+      "learning_rate": 8.593557306971349e-05,
+      "loss": 0.0134108392521739,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.0135,
+      "step": 152,
+      "tokens/total": 4595600,
+      "tokens/train_per_sec_per_gpu": 35.41,
+      "tokens/trainable": 69465
+    },
+    {
+      "epoch": 0.59765625,
+      "grad_norm": 0.5932232141494751,
+      "learning_rate": 8.572411436841618e-05,
+      "loss": 0.03526075929403305,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.03589,
+      "step": 153,
+      "tokens/total": 4626032,
+      "tokens/train_per_sec_per_gpu": 33.9,
+      "tokens/trainable": 69928
+    },
+    {
+      "epoch": 0.6015625,
+      "grad_norm": 0.21574290096759796,
+      "learning_rate": 8.551137711130922e-05,
+      "loss": 0.0023245825432240963,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00233,
+      "step": 154,
+      "tokens/total": 4656352,
+      "tokens/train_per_sec_per_gpu": 36.68,
+      "tokens/trainable": 70395
+    },
+    {
+      "epoch": 0.60546875,
+      "grad_norm": 0.08898661285638809,
+      "learning_rate": 8.529737015125824e-05,
+      "loss": 0.0007986892596818507,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.0008,
+      "step": 155,
+      "tokens/total": 4684672,
+      "tokens/train_per_sec_per_gpu": 35.61,
+      "tokens/trainable": 70829
+    },
+    {
+      "epoch": 0.609375,
+      "grad_norm": 0.1178978830575943,
+      "learning_rate": 8.508210239396639e-05,
+      "loss": 0.0014696650905534625,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00147,
+      "step": 156,
+      "tokens/total": 4715216,
+      "tokens/train_per_sec_per_gpu": 33.9,
+      "tokens/trainable": 71287
+    },
+    {
+      "epoch": 0.61328125,
+      "grad_norm": 0.06257754564285278,
+      "learning_rate": 8.486558279760375e-05,
+      "loss": 0.000776763423345983,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.00078,
+      "step": 157,
+      "tokens/total": 4745376,
+      "tokens/train_per_sec_per_gpu": 35.05,
+      "tokens/trainable": 71749
+    },
+    {
+      "epoch": 0.6171875,
+      "grad_norm": 0.8163263201713562,
+      "learning_rate": 8.464782037243449e-05,
+      "loss": 0.033990029245615005,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.03457,
+      "step": 158,
+      "tokens/total": 4775776,
+      "tokens/train_per_sec_per_gpu": 32.19,
+      "tokens/trainable": 72185
+    },
+    {
+      "epoch": 0.62109375,
+      "grad_norm": 0.014705101028084755,
+      "learning_rate": 8.442882418044202e-05,
+      "loss": 0.00028892228147014976,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.00029,
+      "step": 159,
+      "tokens/total": 4805856,
+      "tokens/train_per_sec_per_gpu": 30.53,
+      "tokens/trainable": 72604
+    },
+    {
+      "epoch": 0.625,
+      "grad_norm": 0.4126265048980713,
+      "learning_rate": 8.420860333495179e-05,
+      "loss": 0.019177839159965515,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01936,
+      "step": 160,
+      "tokens/total": 4836304,
+      "tokens/train_per_sec_per_gpu": 31.58,
+      "tokens/trainable": 73061
+    },
+    {
+      "epoch": 0.62890625,
+      "grad_norm": 0.6193662285804749,
+      "learning_rate": 8.398716700025208e-05,
+      "loss": 0.008987972512841225,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.00903,
+      "step": 161,
+      "tokens/total": 4866544,
+      "tokens/train_per_sec_per_gpu": 29.97,
+      "tokens/trainable": 73474
+    },
+    {
+      "epoch": 0.6328125,
+      "grad_norm": 0.6351296901702881,
+      "learning_rate": 8.376452439121266e-05,
+      "loss": 0.01764395274221897,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.97,
+      "memory/max_allocated (GiB)": 33.97,
+      "ppl": 1.0178,
+      "step": 162,
+      "tokens/total": 4897120,
+      "tokens/train_per_sec_per_gpu": 32.68,
+      "tokens/trainable": 73923
+    },
+    {
+      "epoch": 0.63671875,
+      "grad_norm": 0.19809548556804657,
+      "learning_rate": 8.354068477290124e-05,
+      "loss": 0.002400235738605261,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.0024,
+      "step": 163,
+      "tokens/total": 4927440,
+      "tokens/train_per_sec_per_gpu": 35.37,
+      "tokens/trainable": 74379
+    },
+    {
+      "epoch": 0.640625,
+      "grad_norm": 0.17828311026096344,
+      "learning_rate": 8.331565746019807e-05,
+      "loss": 0.0019721905700862408,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.86,
+      "memory/max_allocated (GiB)": 33.86,
+      "ppl": 1.00197,
+      "step": 164,
+      "tokens/total": 4958032,
+      "tokens/train_per_sec_per_gpu": 33.46,
+      "tokens/trainable": 74830
+    },
+    {
+      "epoch": 0.64453125,
+      "grad_norm": 0.14491626620292664,
+      "learning_rate": 8.308945181740812e-05,
+      "loss": 0.0023866910487413406,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.88,
+      "memory/max_allocated (GiB)": 33.88,
+      "ppl": 1.00239,
+      "step": 165,
+      "tokens/total": 4988368,
+      "tokens/train_per_sec_per_gpu": 31.73,
+      "tokens/trainable": 75262
+    },
+    {
+      "epoch": 0.6484375,
+      "grad_norm": 0.11567407846450806,
+      "learning_rate": 8.286207725787153e-05,
+      "loss": 0.0012497061397880316,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.75,
+      "memory/max_allocated (GiB)": 33.75,
+      "ppl": 1.00125,
+      "step": 166,
+      "tokens/total": 5018672,
+      "tokens/train_per_sec_per_gpu": 31.84,
+      "tokens/trainable": 75703
+    },
+    {
+      "epoch": 0.65234375,
+      "grad_norm": 0.3614877164363861,
+      "learning_rate": 8.263354324357182e-05,
+      "loss": 0.010097509250044823,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.01015,
+      "step": 167,
+      "tokens/total": 5049344,
+      "tokens/train_per_sec_per_gpu": 31.28,
+      "tokens/trainable": 76164
+    },
+    {
+      "epoch": 0.65625,
+      "grad_norm": 0.21049293875694275,
+      "learning_rate": 8.240385928474219e-05,
+      "loss": 0.005174943245947361,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.00519,
+      "step": 168,
+      "tokens/total": 5079856,
+      "tokens/train_per_sec_per_gpu": 34.46,
+      "tokens/trainable": 76623
+    },
+    {
+      "epoch": 0.66015625,
+      "grad_norm": 0.16444046795368195,
+      "learning_rate": 8.217303493946967e-05,
+      "loss": 0.0028489381074905396,
+      "memory/device_reserved (GiB)": 35.82,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00285,
+      "step": 169,
+      "tokens/total": 5110192,
+      "tokens/train_per_sec_per_gpu": 34.39,
+      "tokens/trainable": 77047
+    },
+    {
+      "epoch": 0.6640625,
+      "grad_norm": 0.29660263657569885,
+      "learning_rate": 8.194107981329746e-05,
+      "loss": 0.020341932773590088,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.02055,
+      "step": 170,
+      "tokens/total": 5140592,
+      "tokens/train_per_sec_per_gpu": 34.08,
+      "tokens/trainable": 77496
+    },
+    {
+      "epoch": 0.66796875,
+      "grad_norm": 0.2912541627883911,
+      "learning_rate": 8.170800355882518e-05,
+      "loss": 0.013131296262145042,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.79,
+      "memory/max_allocated (GiB)": 33.79,
+      "ppl": 1.01322,
+      "step": 171,
+      "tokens/total": 5170848,
+      "tokens/train_per_sec_per_gpu": 33.24,
+      "tokens/trainable": 77956
+    },
+    {
+      "epoch": 0.671875,
+      "grad_norm": 0.14199896156787872,
+      "learning_rate": 8.147381587530713e-05,
+      "loss": 0.002253658836707473,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00226,
+      "step": 172,
+      "tokens/total": 5201280,
+      "tokens/train_per_sec_per_gpu": 29.74,
+      "tokens/trainable": 78410
+    },
+    {
+      "epoch": 0.67578125,
+      "grad_norm": 0.6508966684341431,
+      "learning_rate": 8.123852650824877e-05,
+      "loss": 0.01731065846979618,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.01746,
+      "step": 173,
+      "tokens/total": 5231712,
+      "tokens/train_per_sec_per_gpu": 32.99,
+      "tokens/trainable": 78849
+    },
+    {
+      "epoch": 0.6796875,
+      "grad_norm": 0.22890180349349976,
+      "learning_rate": 8.100214524900103e-05,
+      "loss": 0.013438408263027668,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.62,
+      "memory/max_allocated (GiB)": 33.62,
+      "ppl": 1.01353,
+      "step": 174,
+      "tokens/total": 5261440,
+      "tokens/train_per_sec_per_gpu": 30.1,
+      "tokens/trainable": 79266
+    },
+    {
+      "epoch": 0.68359375,
+      "grad_norm": 0.034085191786289215,
+      "learning_rate": 8.076468193435301e-05,
+      "loss": 0.0010016995947808027,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.001,
+      "step": 175,
+      "tokens/total": 5291728,
+      "tokens/train_per_sec_per_gpu": 32.5,
+      "tokens/trainable": 79736
+    },
+    {
+      "epoch": 0.6875,
+      "grad_norm": 0.1346512883901596,
+      "learning_rate": 8.052614644612253e-05,
+      "loss": 0.0026639762800186872,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.93,
+      "memory/max_allocated (GiB)": 33.93,
+      "ppl": 1.00267,
+      "step": 176,
+      "tokens/total": 5322240,
+      "tokens/train_per_sec_per_gpu": 36.14,
+      "tokens/trainable": 80214
+    },
+    {
+      "epoch": 0.69140625,
+      "grad_norm": 0.2848314940929413,
+      "learning_rate": 8.028654871074489e-05,
+      "loss": 0.008686527609825134,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00872,
+      "step": 177,
+      "tokens/total": 5352576,
+      "tokens/train_per_sec_per_gpu": 32.48,
+      "tokens/trainable": 80661
+    },
+    {
+      "epoch": 0.6953125,
+      "grad_norm": 0.037947915494441986,
+      "learning_rate": 8.004589869885986e-05,
+      "loss": 0.0010221066186204553,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.81,
+      "memory/max_allocated (GiB)": 33.81,
+      "ppl": 1.00102,
+      "step": 178,
+      "tokens/total": 5382784,
+      "tokens/train_per_sec_per_gpu": 34.79,
+      "tokens/trainable": 81129
+    },
+    {
+      "epoch": 0.69921875,
+      "grad_norm": 0.1301245242357254,
+      "learning_rate": 7.980420642489674e-05,
+      "loss": 0.0017726586665958166,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00177,
+      "step": 179,
+      "tokens/total": 5412784,
+      "tokens/train_per_sec_per_gpu": 33.04,
+      "tokens/trainable": 81576
+    },
+    {
+      "epoch": 0.703125,
+      "grad_norm": 0.16076700389385223,
+      "learning_rate": 7.95614819466576e-05,
+      "loss": 0.0021704614628106356,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00217,
+      "step": 180,
+      "tokens/total": 5442960,
+      "tokens/train_per_sec_per_gpu": 34.55,
+      "tokens/trainable": 82049
+    },
+    {
+      "epoch": 0.70703125,
+      "grad_norm": 0.5916904807090759,
+      "learning_rate": 7.931773536489872e-05,
+      "loss": 0.018786190077662468,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.01896,
+      "step": 181,
+      "tokens/total": 5473216,
+      "tokens/train_per_sec_per_gpu": 38.83,
+      "tokens/trainable": 82566
+    },
+    {
+      "epoch": 0.7109375,
+      "grad_norm": 0.140871062874794,
+      "learning_rate": 7.907297682291035e-05,
+      "loss": 0.002907736459746957,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.87,
+      "memory/max_allocated (GiB)": 33.87,
+      "ppl": 1.00291,
+      "step": 182,
+      "tokens/total": 5503568,
+      "tokens/train_per_sec_per_gpu": 33.93,
+      "tokens/trainable": 83054
+    },
+    {
+      "epoch": 0.71484375,
+      "grad_norm": 0.3186001181602478,
+      "learning_rate": 7.882721650609442e-05,
+      "loss": 0.002825568662956357,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00283,
+      "step": 183,
+      "tokens/total": 5533760,
+      "tokens/train_per_sec_per_gpu": 32.79,
+      "tokens/trainable": 83501
+    },
+    {
+      "epoch": 0.71875,
+      "grad_norm": 0.10091359168291092,
+      "learning_rate": 7.85804646415409e-05,
+      "loss": 0.000788951525464654,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.89,
+      "memory/max_allocated (GiB)": 33.89,
+      "ppl": 1.00079,
+      "step": 184,
+      "tokens/total": 5564240,
+      "tokens/train_per_sec_per_gpu": 35.52,
+      "tokens/trainable": 84007
+    },
+    {
+      "epoch": 0.72265625,
+      "grad_norm": 0.29195457696914673,
+      "learning_rate": 7.833273149760207e-05,
+      "loss": 0.006313872057944536,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.82,
+      "memory/max_allocated (GiB)": 33.82,
+      "ppl": 1.00633,
+      "step": 185,
+      "tokens/total": 5594448,
+      "tokens/train_per_sec_per_gpu": 35.57,
+      "tokens/trainable": 84491
+    },
+    {
+      "epoch": 0.7265625,
+      "grad_norm": 0.5876129269599915,
+      "learning_rate": 7.808402738346527e-05,
+      "loss": 0.01837407797574997,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.9,
+      "memory/max_allocated (GiB)": 33.9,
+      "ppl": 1.01854,
+      "step": 186,
+      "tokens/total": 5624816,
+      "tokens/train_per_sec_per_gpu": 33.41,
+      "tokens/trainable": 84952
+    },
+    {
+      "epoch": 0.73046875,
+      "grad_norm": 0.4564031660556793,
+      "learning_rate": 7.783436264872382e-05,
+      "loss": 0.00939339492470026,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.84,
+      "memory/max_allocated (GiB)": 33.84,
+      "ppl": 1.00944,
+      "step": 187,
+      "tokens/total": 5655104,
+      "tokens/train_per_sec_per_gpu": 32.93,
+      "tokens/trainable": 85400
+    },
+    {
+      "epoch": 0.734375,
+      "grad_norm": 0.018894854933023453,
+      "learning_rate": 7.758374768294647e-05,
+      "loss": 0.0005394529434852302,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.85,
+      "memory/max_allocated (GiB)": 33.85,
+      "ppl": 1.00054,
+      "step": 188,
+      "tokens/total": 5685360,
+      "tokens/train_per_sec_per_gpu": 33.98,
+      "tokens/trainable": 85889
+    },
+    {
+      "epoch": 0.73828125,
+      "grad_norm": 0.11897671222686768,
+      "learning_rate": 7.733219291524489e-05,
+      "loss": 0.0019742052536457777,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.8,
+      "memory/max_allocated (GiB)": 33.8,
+      "ppl": 1.00198,
+      "step": 189,
+      "tokens/total": 5715648,
+      "tokens/train_per_sec_per_gpu": 36.25,
+      "tokens/trainable": 86396
+    },
+    {
+      "epoch": 0.7421875,
+      "grad_norm": 0.24015508592128754,
+      "learning_rate": 7.707970881383977e-05,
+      "loss": 0.007755734957754612,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.91,
+      "memory/max_allocated (GiB)": 33.91,
+      "ppl": 1.00779,
+      "step": 190,
+      "tokens/total": 5746240,
+      "tokens/train_per_sec_per_gpu": 33.86,
+      "tokens/trainable": 86867
+    },
+    {
+      "epoch": 0.74609375,
+      "grad_norm": 0.2647373378276825,
+      "learning_rate": 7.682630588562518e-05,
+      "loss": 0.017713403329253197,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.99,
+      "memory/max_allocated (GiB)": 33.99,
+      "ppl": 1.01787,
+      "step": 191,
+      "tokens/total": 5776816,
+      "tokens/train_per_sec_per_gpu": 36.16,
+      "tokens/trainable": 87314
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.2803303599357605,
+      "learning_rate": 7.657199467573129e-05,
+      "loss": 0.008597557432949543,
+      "memory/device_reserved (GiB)": 36.02,
+      "memory/max_active (GiB)": 33.83,
+      "memory/max_allocated (GiB)": 33.83,
+      "ppl": 1.00863,
+      "step": 192,
+      "tokens/total": 5807248,
+      "tokens/train_per_sec_per_gpu": 30.38,
+      "tokens/trainable": 87736
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 512,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 32,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.941718578306565e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-192/training_args.bin

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-224/README.md

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

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-224/adapter_config.json

@@ -0,0 +1,48 @@
+{
+  "alora_invocation_tokens": null,
+  "alpha_pattern": {},
+  "arrow_config": null,
+  "auto_mapping": null,
+  "base_model_name_or_path": "/workspace/wave/parent",
+  "bias": "none",
+  "corda_config": null,
+  "ensure_weight_tying": false,
+  "eva_config": null,
+  "exclude_modules": null,
+  "fan_in_fan_out": null,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "layer_replication": null,
+  "layers_pattern": null,
+  "layers_to_transform": null,
+  "loftq_config": {},
+  "lora_alpha": 64,
+  "lora_bias": false,
+  "lora_dropout": 0.05,
+  "lora_ga_config": null,
+  "megatron_config": null,
+  "megatron_core": "megatron.core",
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "peft_version": "0.19.1",
+  "qalora_group_size": 16,
+  "r": 32,
+  "rank_pattern": {},
+  "revision": null,
+  "target_modules": [
+    "k_proj",
+    "o_proj",
+    "down_proj",
+    "up_proj",
+    "q_proj",
+    "v_proj",
+    "gate_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__charter0p2/training/checkpoints/checkpoint-224/adapter_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d107b6d944d7d322dfcfa139e7e137132a50659242bcb73f462c737f806835c5
+size 547777976

aft_wave_v2/control_matched__charter0p2/training/checkpoints/checkpoint-224/chat_template.jinja

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