Pilot model commit

config.json

@@ -0,0 +1,33 @@
+{
+  "architectures": [
+    "CharformerCanineForMaskedLM"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 57344,
+  "downsampling_rate": 4,
+  "eos_token_id": 57345,
+  "gbst_dim": 768,
+  "gbst_downsample_factor": 4,
+  "gbst_max_block_size": 8,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "local_transformer_stride": 128,
+  "mask_token_id": 1,
+  "max_position_embeddings": 8192,
+  "max_span_length": 10,
+  "model_type": "canine",
+  "num_attention_heads": 12,
+  "num_hash_buckets": 16384,
+  "num_hash_functions": 8,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "torch_dtype": "float32",
+  "transformers_version": "4.48.3",
+  "type_vocab_size": 16,
+  "upsampling_kernel_size": 4,
+  "vocab_size": 38
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab0a0fafb1d3cb664a6fc231ad7ea39b139840768c445f2d3492a35f3f116806
+size 672028372

optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29fcfe20c070ae55328034692cabf27638362d3bd4aa38dd03d32611e8895401
+size 1343934394

rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c1f1480f6b13bfd62e2d168a109fa44c1b8dc8a166c392b4909dd01cac470257
+size 14308

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e23bbca716660e478985d3ba5d365753812994c8f83e01527d8d3d4766052f4
+size 1064

trainer_state.json

@@ -0,0 +1,2328 @@
+{
+  "best_metric": 0.9716248512268066,
+  "best_model_checkpoint": "/scratch/gpfs/BG11/char-model/char-model-linear-30-V2/checkpoint-244800",
+  "epoch": 30.0,
+  "eval_steps": 1600,
+  "global_step": 245640,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.19540791402051783,
+      "grad_norm": 1.4876313209533691,
+      "learning_rate": 8e-05,
+      "loss": 2.5789,
+      "step": 1600
+    },
+    {
+      "epoch": 0.19540791402051783,
+      "eval_loss": 2.3971974849700928,
+      "eval_runtime": 114.1562,
+      "eval_samples_per_second": 255.019,
+      "eval_steps_per_second": 7.972,
+      "step": 1600
+    },
+    {
+      "epoch": 0.39081582804103565,
+      "grad_norm": 0.6994622349739075,
+      "learning_rate": 0.00016,
+      "loss": 2.2769,
+      "step": 3200
+    },
+    {
+      "epoch": 0.39081582804103565,
+      "eval_loss": 2.1549973487854004,
+      "eval_runtime": 114.8328,
+      "eval_samples_per_second": 253.516,
+      "eval_steps_per_second": 7.925,
+      "step": 3200
+    },
+    {
+      "epoch": 0.5862237420615535,
+      "grad_norm": 0.6388512849807739,
+      "learning_rate": 0.00019999459110993603,
+      "loss": 2.093,
+      "step": 4800
+    },
+    {
+      "epoch": 0.5862237420615535,
+      "eval_loss": 1.955593228340149,
+      "eval_runtime": 114.1957,
+      "eval_samples_per_second": 254.931,
+      "eval_steps_per_second": 7.969,
+      "step": 4800
+    },
+    {
+      "epoch": 0.7816316560820713,
+      "grad_norm": 0.5748873353004456,
+      "learning_rate": 0.00019995132350009195,
+      "loss": 1.9338,
+      "step": 6400
+    },
+    {
+      "epoch": 0.7816316560820713,
+      "eval_loss": 1.803764820098877,
+      "eval_runtime": 114.1068,
+      "eval_samples_per_second": 255.129,
+      "eval_steps_per_second": 7.975,
+      "step": 6400
+    },
+    {
+      "epoch": 0.9770395701025891,
+      "grad_norm": 0.5466416478157043,
+      "learning_rate": 0.0001998648070022771,
+      "loss": 1.83,
+      "step": 8000
+    },
+    {
+      "epoch": 0.9770395701025891,
+      "eval_loss": 1.705185890197754,
+      "eval_runtime": 114.5361,
+      "eval_samples_per_second": 254.173,
+      "eval_steps_per_second": 7.945,
+      "step": 8000
+    },
+    {
+      "epoch": 1.172447484123107,
+      "grad_norm": 0.5952168107032776,
+      "learning_rate": 0.00019973507905213703,
+      "loss": 1.7599,
+      "step": 9600
+    },
+    {
+      "epoch": 1.172447484123107,
+      "eval_loss": 1.640653133392334,
+      "eval_runtime": 114.2365,
+      "eval_samples_per_second": 254.84,
+      "eval_steps_per_second": 7.966,
+      "step": 9600
+    },
+    {
+      "epoch": 1.3678553981436248,
+      "grad_norm": 0.5351431965827942,
+      "learning_rate": 0.00019956219578289123,
+      "loss": 1.7148,
+      "step": 11200
+    },
+    {
+      "epoch": 1.3678553981436248,
+      "eval_loss": 1.593337059020996,
+      "eval_runtime": 114.2436,
+      "eval_samples_per_second": 254.824,
+      "eval_steps_per_second": 7.965,
+      "step": 11200
+    },
+    {
+      "epoch": 1.5632633121641426,
+      "grad_norm": 0.5703535079956055,
+      "learning_rate": 0.0001993462320010443,
+      "loss": 1.6811,
+      "step": 12800
+    },
+    {
+      "epoch": 1.5632633121641426,
+      "eval_loss": 1.563097357749939,
+      "eval_runtime": 114.0333,
+      "eval_samples_per_second": 255.294,
+      "eval_steps_per_second": 7.98,
+      "step": 12800
+    },
+    {
+      "epoch": 1.7586712261846604,
+      "grad_norm": 0.5037602186203003,
+      "learning_rate": 0.00019908728115401733,
+      "loss": 1.6543,
+      "step": 14400
+    },
+    {
+      "epoch": 1.7586712261846604,
+      "eval_loss": 1.532173752784729,
+      "eval_runtime": 114.2165,
+      "eval_samples_per_second": 254.884,
+      "eval_steps_per_second": 7.967,
+      "step": 14400
+    },
+    {
+      "epoch": 1.9540791402051783,
+      "grad_norm": 0.5285059213638306,
+      "learning_rate": 0.00019878545528971298,
+      "loss": 1.6323,
+      "step": 16000
+    },
+    {
+      "epoch": 1.9540791402051783,
+      "eval_loss": 1.5123432874679565,
+      "eval_runtime": 114.0419,
+      "eval_samples_per_second": 255.275,
+      "eval_steps_per_second": 7.98,
+      "step": 16000
+    },
+    {
+      "epoch": 2.149487054225696,
+      "grad_norm": 1.3733247518539429,
+      "learning_rate": 0.00019844088500803263,
+      "loss": 1.6121,
+      "step": 17600
+    },
+    {
+      "epoch": 2.149487054225696,
+      "eval_loss": 1.4912608861923218,
+      "eval_runtime": 114.2353,
+      "eval_samples_per_second": 254.842,
+      "eval_steps_per_second": 7.966,
+      "step": 17600
+    },
+    {
+      "epoch": 2.344894968246214,
+      "grad_norm": 0.5754392147064209,
+      "learning_rate": 0.00019805371940436587,
+      "loss": 1.5963,
+      "step": 19200
+    },
+    {
+      "epoch": 2.344894968246214,
+      "eval_loss": 1.4811656475067139,
+      "eval_runtime": 113.8361,
+      "eval_samples_per_second": 255.736,
+      "eval_steps_per_second": 7.994,
+      "step": 19200
+    },
+    {
+      "epoch": 2.5403028822667317,
+      "grad_norm": 0.5134813189506531,
+      "learning_rate": 0.00019762412600507676,
+      "loss": 1.5805,
+      "step": 20800
+    },
+    {
+      "epoch": 2.5403028822667317,
+      "eval_loss": 1.4569112062454224,
+      "eval_runtime": 113.8761,
+      "eval_samples_per_second": 255.646,
+      "eval_steps_per_second": 7.991,
+      "step": 20800
+    },
+    {
+      "epoch": 2.7357107962872496,
+      "grad_norm": 0.5872719287872314,
+      "learning_rate": 0.0001971522906950156,
+      "loss": 1.5672,
+      "step": 22400
+    },
+    {
+      "epoch": 2.7357107962872496,
+      "eval_loss": 1.4468671083450317,
+      "eval_runtime": 114.3556,
+      "eval_samples_per_second": 254.574,
+      "eval_steps_per_second": 7.958,
+      "step": 22400
+    },
+    {
+      "epoch": 2.9311187103077674,
+      "grad_norm": 0.5536867380142212,
+      "learning_rate": 0.0001966384176370864,
+      "loss": 1.5522,
+      "step": 24000
+    },
+    {
+      "epoch": 2.9311187103077674,
+      "eval_loss": 1.4352666139602661,
+      "eval_runtime": 114.2848,
+      "eval_samples_per_second": 254.732,
+      "eval_steps_per_second": 7.963,
+      "step": 24000
+    },
+    {
+      "epoch": 3.126526624328285,
+      "grad_norm": 0.5568290948867798,
+      "learning_rate": 0.00019608272918390576,
+      "loss": 1.544,
+      "step": 25600
+    },
+    {
+      "epoch": 3.126526624328285,
+      "eval_loss": 1.4248650074005127,
+      "eval_runtime": 114.5178,
+      "eval_samples_per_second": 254.214,
+      "eval_steps_per_second": 7.946,
+      "step": 25600
+    },
+    {
+      "epoch": 3.321934538348803,
+      "grad_norm": 0.643578052520752,
+      "learning_rate": 0.00019548546578159097,
+      "loss": 1.5315,
+      "step": 27200
+    },
+    {
+      "epoch": 3.321934538348803,
+      "eval_loss": 1.4156839847564697,
+      "eval_runtime": 114.7135,
+      "eval_samples_per_second": 253.78,
+      "eval_steps_per_second": 7.933,
+      "step": 27200
+    },
+    {
+      "epoch": 3.517342452369321,
+      "grad_norm": 0.5677089691162109,
+      "learning_rate": 0.00019484688586571922,
+      "loss": 1.5212,
+      "step": 28800
+    },
+    {
+      "epoch": 3.517342452369321,
+      "eval_loss": 1.3995802402496338,
+      "eval_runtime": 114.5983,
+      "eval_samples_per_second": 254.035,
+      "eval_steps_per_second": 7.941,
+      "step": 28800
+    },
+    {
+      "epoch": 3.7127503663898387,
+      "grad_norm": 0.6246566772460938,
+      "learning_rate": 0.00019416726574950254,
+      "loss": 1.5141,
+      "step": 30400
+    },
+    {
+      "epoch": 3.7127503663898387,
+      "eval_loss": 1.390926718711853,
+      "eval_runtime": 114.3382,
+      "eval_samples_per_second": 254.613,
+      "eval_steps_per_second": 7.959,
+      "step": 30400
+    },
+    {
+      "epoch": 3.9081582804103565,
+      "grad_norm": 0.5574250221252441,
+      "learning_rate": 0.0001934468995042272,
+      "loss": 1.5031,
+      "step": 32000
+    },
+    {
+      "epoch": 3.9081582804103565,
+      "eval_loss": 1.3884520530700684,
+      "eval_runtime": 114.5913,
+      "eval_samples_per_second": 254.051,
+      "eval_steps_per_second": 7.941,
+      "step": 32000
+    },
+    {
+      "epoch": 4.103566194430875,
+      "grad_norm": 0.6128900051116943,
+      "learning_rate": 0.00019268609883200935,
+      "loss": 1.4957,
+      "step": 33600
+    },
+    {
+      "epoch": 4.103566194430875,
+      "eval_loss": 1.3740371465682983,
+      "eval_runtime": 114.6971,
+      "eval_samples_per_second": 253.816,
+      "eval_steps_per_second": 7.934,
+      "step": 33600
+    },
+    {
+      "epoch": 4.298974108451392,
+      "grad_norm": 0.5937383770942688,
+      "learning_rate": 0.00019188519293092138,
+      "loss": 1.4874,
+      "step": 35200
+    },
+    {
+      "epoch": 4.298974108451392,
+      "eval_loss": 1.3618671894073486,
+      "eval_runtime": 114.3011,
+      "eval_samples_per_second": 254.696,
+      "eval_steps_per_second": 7.961,
+      "step": 35200
+    },
+    {
+      "epoch": 4.49438202247191,
+      "grad_norm": 0.5859814286231995,
+      "learning_rate": 0.00019104452835254848,
+      "loss": 1.4788,
+      "step": 36800
+    },
+    {
+      "epoch": 4.49438202247191,
+      "eval_loss": 1.3512194156646729,
+      "eval_runtime": 114.1714,
+      "eval_samples_per_second": 254.985,
+      "eval_steps_per_second": 7.97,
+      "step": 36800
+    },
+    {
+      "epoch": 4.689789936492428,
+      "grad_norm": 0.6147826910018921,
+      "learning_rate": 0.00019016446885203558,
+      "loss": 1.4728,
+      "step": 38400
+    },
+    {
+      "epoch": 4.689789936492428,
+      "eval_loss": 1.346225619316101,
+      "eval_runtime": 114.1824,
+      "eval_samples_per_second": 254.961,
+      "eval_steps_per_second": 7.97,
+      "step": 38400
+    },
+    {
+      "epoch": 4.885197850512946,
+      "grad_norm": 0.6011592745780945,
+      "learning_rate": 0.00018924539523069092,
+      "loss": 1.4642,
+      "step": 40000
+    },
+    {
+      "epoch": 4.885197850512946,
+      "eval_loss": 1.3416181802749634,
+      "eval_runtime": 114.2305,
+      "eval_samples_per_second": 254.853,
+      "eval_steps_per_second": 7.966,
+      "step": 40000
+    },
+    {
+      "epoch": 5.0806057645334635,
+      "grad_norm": 0.5911765098571777,
+      "learning_rate": 0.00018828770517121334,
+      "loss": 1.4573,
+      "step": 41600
+    },
+    {
+      "epoch": 5.0806057645334635,
+      "eval_loss": 1.3351951837539673,
+      "eval_runtime": 113.6497,
+      "eval_samples_per_second": 256.156,
+      "eval_steps_per_second": 8.007,
+      "step": 41600
+    },
+    {
+      "epoch": 5.276013678553982,
+      "grad_norm": 0.6570513844490051,
+      "learning_rate": 0.00018729181306561534,
+      "loss": 1.4516,
+      "step": 43200
+    },
+    {
+      "epoch": 5.276013678553982,
+      "eval_loss": 1.3280487060546875,
+      "eval_runtime": 113.3621,
+      "eval_samples_per_second": 256.806,
+      "eval_steps_per_second": 8.027,
+      "step": 43200
+    },
+    {
+      "epoch": 5.471421592574499,
+      "grad_norm": 0.6237516403198242,
+      "learning_rate": 0.00018625814983591572,
+      "loss": 1.445,
+      "step": 44800
+    },
+    {
+      "epoch": 5.471421592574499,
+      "eval_loss": 1.3218315839767456,
+      "eval_runtime": 113.4,
+      "eval_samples_per_second": 256.72,
+      "eval_steps_per_second": 8.025,
+      "step": 44800
+    },
+    {
+      "epoch": 5.666829506595017,
+      "grad_norm": 0.6138856410980225,
+      "learning_rate": 0.00018518716274767993,
+      "loss": 1.4403,
+      "step": 46400
+    },
+    {
+      "epoch": 5.666829506595017,
+      "eval_loss": 1.3159008026123047,
+      "eval_runtime": 113.486,
+      "eval_samples_per_second": 256.525,
+      "eval_steps_per_second": 8.019,
+      "step": 46400
+    },
+    {
+      "epoch": 5.862237420615535,
+      "grad_norm": 0.6461225152015686,
+      "learning_rate": 0.0001840793152164884,
+      "loss": 1.4338,
+      "step": 48000
+    },
+    {
+      "epoch": 5.862237420615535,
+      "eval_loss": 1.307564377784729,
+      "eval_runtime": 113.2007,
+      "eval_samples_per_second": 257.171,
+      "eval_steps_per_second": 8.039,
+      "step": 48000
+    },
+    {
+      "epoch": 6.057645334636053,
+      "grad_norm": 0.6271090507507324,
+      "learning_rate": 0.0001829350866074169,
+      "loss": 1.4241,
+      "step": 49600
+    },
+    {
+      "epoch": 6.057645334636053,
+      "eval_loss": 1.3019992113113403,
+      "eval_runtime": 113.8204,
+      "eval_samples_per_second": 255.771,
+      "eval_steps_per_second": 7.995,
+      "step": 49600
+    },
+    {
+      "epoch": 6.25305324865657,
+      "grad_norm": 0.6223775148391724,
+      "learning_rate": 0.0001817549720276156,
+      "loss": 1.4204,
+      "step": 51200
+    },
+    {
+      "epoch": 6.25305324865657,
+      "eval_loss": 1.2931100130081177,
+      "eval_runtime": 113.6749,
+      "eval_samples_per_second": 256.099,
+      "eval_steps_per_second": 8.005,
+      "step": 51200
+    },
+    {
+      "epoch": 6.448461162677089,
+      "grad_norm": 0.652824342250824,
+      "learning_rate": 0.00018053948211207624,
+      "loss": 1.4125,
+      "step": 52800
+    },
+    {
+      "epoch": 6.448461162677089,
+      "eval_loss": 1.2813327312469482,
+      "eval_runtime": 113.1715,
+      "eval_samples_per_second": 257.238,
+      "eval_steps_per_second": 8.041,
+      "step": 52800
+    },
+    {
+      "epoch": 6.643869076697606,
+      "grad_norm": 0.6772754788398743,
+      "learning_rate": 0.0001792891428026808,
+      "loss": 1.406,
+      "step": 54400
+    },
+    {
+      "epoch": 6.643869076697606,
+      "eval_loss": 1.2822933197021484,
+      "eval_runtime": 113.3373,
+      "eval_samples_per_second": 256.862,
+      "eval_steps_per_second": 8.029,
+      "step": 54400
+    },
+    {
+      "epoch": 6.839276990718124,
+      "grad_norm": 0.6514716148376465,
+      "learning_rate": 0.0001780044951206266,
+      "loss": 1.3993,
+      "step": 56000
+    },
+    {
+      "epoch": 6.839276990718124,
+      "eval_loss": 1.2731530666351318,
+      "eval_runtime": 113.4336,
+      "eval_samples_per_second": 256.643,
+      "eval_steps_per_second": 8.022,
+      "step": 56000
+    },
+    {
+      "epoch": 7.034684904738642,
+      "grad_norm": 0.7061969041824341,
+      "learning_rate": 0.00017668609493232685,
+      "loss": 1.3954,
+      "step": 57600
+    },
+    {
+      "epoch": 7.034684904738642,
+      "eval_loss": 1.2732563018798828,
+      "eval_runtime": 113.5204,
+      "eval_samples_per_second": 256.447,
+      "eval_steps_per_second": 8.016,
+      "step": 57600
+    },
+    {
+      "epoch": 7.23009281875916,
+      "grad_norm": 0.6283796429634094,
+      "learning_rate": 0.00017533451270888733,
+      "loss": 1.3885,
+      "step": 59200
+    },
+    {
+      "epoch": 7.23009281875916,
+      "eval_loss": 1.2614556550979614,
+      "eval_runtime": 113.9439,
+      "eval_samples_per_second": 255.494,
+      "eval_steps_per_second": 7.986,
+      "step": 59200
+    },
+    {
+      "epoch": 7.425500732779677,
+      "grad_norm": 0.701786994934082,
+      "learning_rate": 0.00017395033327926394,
+      "loss": 1.3839,
+      "step": 60800
+    },
+    {
+      "epoch": 7.425500732779677,
+      "eval_loss": 1.2564245462417603,
+      "eval_runtime": 113.6613,
+      "eval_samples_per_second": 256.129,
+      "eval_steps_per_second": 8.006,
+      "step": 60800
+    },
+    {
+      "epoch": 7.620908646800196,
+      "grad_norm": 0.6613229513168335,
+      "learning_rate": 0.0001725341555772075,
+      "loss": 1.379,
+      "step": 62400
+    },
+    {
+      "epoch": 7.620908646800196,
+      "eval_loss": 1.248888373374939,
+      "eval_runtime": 113.7924,
+      "eval_samples_per_second": 255.834,
+      "eval_steps_per_second": 7.997,
+      "step": 62400
+    },
+    {
+      "epoch": 7.816316560820713,
+      "grad_norm": 0.6857264041900635,
+      "learning_rate": 0.00017108659238210543,
+      "loss": 1.3737,
+      "step": 64000
+    },
+    {
+      "epoch": 7.816316560820713,
+      "eval_loss": 1.2474324703216553,
+      "eval_runtime": 114.1649,
+      "eval_samples_per_second": 255.0,
+      "eval_steps_per_second": 7.971,
+      "step": 64000
+    },
+    {
+      "epoch": 8.01172447484123,
+      "grad_norm": 0.6922156810760498,
+      "learning_rate": 0.00016960827005383234,
+      "loss": 1.3686,
+      "step": 65600
+    },
+    {
+      "epoch": 8.01172447484123,
+      "eval_loss": 1.2473887205123901,
+      "eval_runtime": 113.4996,
+      "eval_samples_per_second": 256.494,
+      "eval_steps_per_second": 8.018,
+      "step": 65600
+    },
+    {
+      "epoch": 8.20713238886175,
+      "grad_norm": 0.6817461252212524,
+      "learning_rate": 0.00016809982826172446,
+      "loss": 1.3603,
+      "step": 67200
+    },
+    {
+      "epoch": 8.20713238886175,
+      "eval_loss": 1.2385778427124023,
+      "eval_runtime": 113.3075,
+      "eval_samples_per_second": 256.929,
+      "eval_steps_per_second": 8.031,
+      "step": 67200
+    },
+    {
+      "epoch": 8.402540302882267,
+      "grad_norm": 0.6611928343772888,
+      "learning_rate": 0.00016656191970779508,
+      "loss": 1.3582,
+      "step": 68800
+    },
+    {
+      "epoch": 8.402540302882267,
+      "eval_loss": 1.232322096824646,
+      "eval_runtime": 113.1998,
+      "eval_samples_per_second": 257.174,
+      "eval_steps_per_second": 8.039,
+      "step": 68800
+    },
+    {
+      "epoch": 8.597948216902784,
+      "grad_norm": 0.7477986812591553,
+      "learning_rate": 0.0001649952098443106,
+      "loss": 1.3533,
+      "step": 70400
+    },
+    {
+      "epoch": 8.597948216902784,
+      "eval_loss": 1.2227420806884766,
+      "eval_runtime": 113.2449,
+      "eval_samples_per_second": 257.071,
+      "eval_steps_per_second": 8.036,
+      "step": 70400
+    },
+    {
+      "epoch": 8.793356130923302,
+      "grad_norm": 0.6740342378616333,
+      "learning_rate": 0.00016340037658584987,
+      "loss": 1.3479,
+      "step": 72000
+    },
+    {
+      "epoch": 8.793356130923302,
+      "eval_loss": 1.2295479774475098,
+      "eval_runtime": 113.3877,
+      "eval_samples_per_second": 256.747,
+      "eval_steps_per_second": 8.026,
+      "step": 72000
+    },
+    {
+      "epoch": 8.98876404494382,
+      "grad_norm": 0.6922385096549988,
+      "learning_rate": 0.00016177811001597065,
+      "loss": 1.3436,
+      "step": 73600
+    },
+    {
+      "epoch": 8.98876404494382,
+      "eval_loss": 1.211731195449829,
+      "eval_runtime": 113.5374,
+      "eval_samples_per_second": 256.409,
+      "eval_steps_per_second": 8.015,
+      "step": 73600
+    },
+    {
+      "epoch": 9.184171958964338,
+      "grad_norm": 0.711102306842804,
+      "learning_rate": 0.00016012911208861095,
+      "loss": 1.3379,
+      "step": 75200
+    },
+    {
+      "epoch": 9.184171958964338,
+      "eval_loss": 1.2126644849777222,
+      "eval_runtime": 113.5611,
+      "eval_samples_per_second": 256.355,
+      "eval_steps_per_second": 8.013,
+      "step": 75200
+    },
+    {
+      "epoch": 9.379579872984856,
+      "grad_norm": 0.6939982175827026,
+      "learning_rate": 0.00015845409632435383,
+      "loss": 1.3308,
+      "step": 76800
+    },
+    {
+      "epoch": 9.379579872984856,
+      "eval_loss": 1.2125294208526611,
+      "eval_runtime": 113.3442,
+      "eval_samples_per_second": 256.846,
+      "eval_steps_per_second": 8.029,
+      "step": 76800
+    },
+    {
+      "epoch": 9.574987787005373,
+      "grad_norm": 0.7124617695808411,
+      "learning_rate": 0.0001567537875016875,
+      "loss": 1.328,
+      "step": 78400
+    },
+    {
+      "epoch": 9.574987787005373,
+      "eval_loss": 1.2044532299041748,
+      "eval_runtime": 113.4468,
+      "eval_samples_per_second": 256.614,
+      "eval_steps_per_second": 8.021,
+      "step": 78400
+    },
+    {
+      "epoch": 9.770395701025892,
+      "grad_norm": 0.7330592274665833,
+      "learning_rate": 0.00015502892134339392,
+      "loss": 1.3218,
+      "step": 80000
+    },
+    {
+      "epoch": 9.770395701025892,
+      "eval_loss": 1.1967058181762695,
+      "eval_runtime": 113.2601,
+      "eval_samples_per_second": 257.037,
+      "eval_steps_per_second": 8.035,
+      "step": 80000
+    },
+    {
+      "epoch": 9.96580361504641,
+      "grad_norm": 0.6967942714691162,
+      "learning_rate": 0.00015328024419820202,
+      "loss": 1.3183,
+      "step": 81600
+    },
+    {
+      "epoch": 9.96580361504641,
+      "eval_loss": 1.1870883703231812,
+      "eval_runtime": 113.4792,
+      "eval_samples_per_second": 256.541,
+      "eval_steps_per_second": 8.019,
+      "step": 81600
+    },
+    {
+      "epoch": 10.161211529066927,
+      "grad_norm": 0.7507591843605042,
+      "learning_rate": 0.00015150851271784278,
+      "loss": 1.3141,
+      "step": 83200
+    },
+    {
+      "epoch": 10.161211529066927,
+      "eval_loss": 1.1836739778518677,
+      "eval_runtime": 113.0477,
+      "eval_samples_per_second": 257.52,
+      "eval_steps_per_second": 8.05,
+      "step": 83200
+    },
+    {
+      "epoch": 10.356619443087444,
+      "grad_norm": 0.700063169002533,
+      "learning_rate": 0.00014971449352964674,
+      "loss": 1.3069,
+      "step": 84800
+    },
+    {
+      "epoch": 10.356619443087444,
+      "eval_loss": 1.1819714307785034,
+      "eval_runtime": 113.2705,
+      "eval_samples_per_second": 257.013,
+      "eval_steps_per_second": 8.034,
+      "step": 84800
+    },
+    {
+      "epoch": 10.552027357107963,
+      "grad_norm": 0.7527259588241577,
+      "learning_rate": 0.00014789896290482433,
+      "loss": 1.3023,
+      "step": 86400
+    },
+    {
+      "epoch": 10.552027357107963,
+      "eval_loss": 1.180099606513977,
+      "eval_runtime": 113.5043,
+      "eval_samples_per_second": 256.484,
+      "eval_steps_per_second": 8.017,
+      "step": 86400
+    },
+    {
+      "epoch": 10.74743527112848,
+      "grad_norm": 0.7384325861930847,
+      "learning_rate": 0.00014606270642257408,
+      "loss": 1.3004,
+      "step": 88000
+    },
+    {
+      "epoch": 10.74743527112848,
+      "eval_loss": 1.1687953472137451,
+      "eval_runtime": 113.2405,
+      "eval_samples_per_second": 257.081,
+      "eval_steps_per_second": 8.036,
+      "step": 88000
+    },
+    {
+      "epoch": 10.942843185148998,
+      "grad_norm": 0.7243154644966125,
+      "learning_rate": 0.00014420651863016263,
+      "loss": 1.2952,
+      "step": 89600
+    },
+    {
+      "epoch": 10.942843185148998,
+      "eval_loss": 1.1710914373397827,
+      "eval_runtime": 113.1239,
+      "eval_samples_per_second": 257.346,
+      "eval_steps_per_second": 8.044,
+      "step": 89600
+    },
+    {
+      "epoch": 11.138251099169516,
+      "grad_norm": 0.7848924994468689,
+      "learning_rate": 0.0001423312026991247,
+      "loss": 1.2897,
+      "step": 91200
+    },
+    {
+      "epoch": 11.138251099169516,
+      "eval_loss": 1.169708013534546,
+      "eval_runtime": 113.4134,
+      "eval_samples_per_second": 256.689,
+      "eval_steps_per_second": 8.024,
+      "step": 91200
+    },
+    {
+      "epoch": 11.333659013190035,
+      "grad_norm": 0.7706081867218018,
+      "learning_rate": 0.00014043757007773094,
+      "loss": 1.2844,
+      "step": 92800
+    },
+    {
+      "epoch": 11.333659013190035,
+      "eval_loss": 1.1610310077667236,
+      "eval_runtime": 113.6197,
+      "eval_samples_per_second": 256.223,
+      "eval_steps_per_second": 8.009,
+      "step": 92800
+    },
+    {
+      "epoch": 11.529066927210552,
+      "grad_norm": 0.7377546429634094,
+      "learning_rate": 0.00013852644013987489,
+      "loss": 1.2803,
+      "step": 94400
+    },
+    {
+      "epoch": 11.529066927210552,
+      "eval_loss": 1.1572864055633545,
+      "eval_runtime": 113.4329,
+      "eval_samples_per_second": 256.645,
+      "eval_steps_per_second": 8.022,
+      "step": 94400
+    },
+    {
+      "epoch": 11.72447484123107,
+      "grad_norm": 0.7930458188056946,
+      "learning_rate": 0.00013659863983053026,
+      "loss": 1.2777,
+      "step": 96000
+    },
+    {
+      "epoch": 11.72447484123107,
+      "eval_loss": 1.1565219163894653,
+      "eval_runtime": 113.3539,
+      "eval_samples_per_second": 256.824,
+      "eval_steps_per_second": 8.028,
+      "step": 96000
+    },
+    {
+      "epoch": 11.919882755251587,
+      "grad_norm": 0.7491051554679871,
+      "learning_rate": 0.00013465500330793215,
+      "loss": 1.2734,
+      "step": 97600
+    },
+    {
+      "epoch": 11.919882755251587,
+      "eval_loss": 1.148636817932129,
+      "eval_runtime": 113.4666,
+      "eval_samples_per_second": 256.569,
+      "eval_steps_per_second": 8.02,
+      "step": 97600
+    },
+    {
+      "epoch": 12.115290669272106,
+      "grad_norm": 0.765771746635437,
+      "learning_rate": 0.0001326963715826373,
+      "loss": 1.2688,
+      "step": 99200
+    },
+    {
+      "epoch": 12.115290669272106,
+      "eval_loss": 1.1487972736358643,
+      "eval_runtime": 113.1177,
+      "eval_samples_per_second": 257.36,
+      "eval_steps_per_second": 8.045,
+      "step": 99200
+    },
+    {
+      "epoch": 12.310698583292623,
+      "grad_norm": 0.7452722191810608,
+      "learning_rate": 0.00013072359215361915,
+      "loss": 1.2646,
+      "step": 100800
+    },
+    {
+      "epoch": 12.310698583292623,
+      "eval_loss": 1.1423231363296509,
+      "eval_runtime": 113.0385,
+      "eval_samples_per_second": 257.541,
+      "eval_steps_per_second": 8.05,
+      "step": 100800
+    },
+    {
+      "epoch": 12.50610649731314,
+      "grad_norm": 0.8033236265182495,
+      "learning_rate": 0.00012873751864155556,
+      "loss": 1.2604,
+      "step": 102400
+    },
+    {
+      "epoch": 12.50610649731314,
+      "eval_loss": 1.135223627090454,
+      "eval_runtime": 113.5509,
+      "eval_samples_per_second": 256.378,
+      "eval_steps_per_second": 8.014,
+      "step": 102400
+    },
+    {
+      "epoch": 12.701514411333658,
+      "grad_norm": 0.7842743396759033,
+      "learning_rate": 0.0001267390104194675,
+      "loss": 1.2566,
+      "step": 104000
+    },
+    {
+      "epoch": 12.701514411333658,
+      "eval_loss": 1.1383661031723022,
+      "eval_runtime": 113.3823,
+      "eval_samples_per_second": 256.76,
+      "eval_steps_per_second": 8.026,
+      "step": 104000
+    },
+    {
+      "epoch": 12.896922325354177,
+      "grad_norm": 0.7673171758651733,
+      "learning_rate": 0.00012472893224086873,
+      "loss": 1.2534,
+      "step": 105600
+    },
+    {
+      "epoch": 12.896922325354177,
+      "eval_loss": 1.1316802501678467,
+      "eval_runtime": 113.8092,
+      "eval_samples_per_second": 255.797,
+      "eval_steps_per_second": 7.996,
+      "step": 105600
+    },
+    {
+      "epoch": 13.092330239374695,
+      "grad_norm": 4.324079513549805,
+      "learning_rate": 0.00012270815386558753,
+      "loss": 1.2495,
+      "step": 107200
+    },
+    {
+      "epoch": 13.092330239374695,
+      "eval_loss": 1.1293587684631348,
+      "eval_runtime": 113.1334,
+      "eval_samples_per_second": 257.324,
+      "eval_steps_per_second": 8.044,
+      "step": 107200
+    },
+    {
+      "epoch": 13.287738153395212,
+      "grad_norm": 0.8010779023170471,
+      "learning_rate": 0.00012067754968342186,
+      "loss": 1.2429,
+      "step": 108800
+    },
+    {
+      "epoch": 13.287738153395212,
+      "eval_loss": 1.1205388307571411,
+      "eval_runtime": 113.0887,
+      "eval_samples_per_second": 257.426,
+      "eval_steps_per_second": 8.047,
+      "step": 108800
+    },
+    {
+      "epoch": 13.48314606741573,
+      "grad_norm": 0.7941340804100037,
+      "learning_rate": 0.00011863799833579153,
+      "loss": 1.2403,
+      "step": 110400
+    },
+    {
+      "epoch": 13.48314606741573,
+      "eval_loss": 1.1161994934082031,
+      "eval_runtime": 112.9454,
+      "eval_samples_per_second": 257.753,
+      "eval_steps_per_second": 8.057,
+      "step": 110400
+    },
+    {
+      "epoch": 13.678553981436249,
+      "grad_norm": 0.746963381767273,
+      "learning_rate": 0.00011659038233555033,
+      "loss": 1.2386,
+      "step": 112000
+    },
+    {
+      "epoch": 13.678553981436249,
+      "eval_loss": 1.1152887344360352,
+      "eval_runtime": 113.0036,
+      "eval_samples_per_second": 257.62,
+      "eval_steps_per_second": 8.053,
+      "step": 112000
+    },
+    {
+      "epoch": 13.873961895456766,
+      "grad_norm": 0.7765190601348877,
+      "learning_rate": 0.00011453558768512322,
+      "loss": 1.2332,
+      "step": 113600
+    },
+    {
+      "epoch": 13.873961895456766,
+      "eval_loss": 1.1129028797149658,
+      "eval_runtime": 113.3982,
+      "eval_samples_per_second": 256.724,
+      "eval_steps_per_second": 8.025,
+      "step": 113600
+    },
+    {
+      "epoch": 14.069369809477283,
+      "grad_norm": 0.8052350282669067,
+      "learning_rate": 0.00011247450349313363,
+      "loss": 1.2326,
+      "step": 115200
+    },
+    {
+      "epoch": 14.069369809477283,
+      "eval_loss": 1.107361078262329,
+      "eval_runtime": 113.042,
+      "eval_samples_per_second": 257.533,
+      "eval_steps_per_second": 8.05,
+      "step": 115200
+    },
+    {
+      "epoch": 14.2647777234978,
+      "grad_norm": 0.7904307246208191,
+      "learning_rate": 0.00011040802158968633,
+      "loss": 1.2251,
+      "step": 116800
+    },
+    {
+      "epoch": 14.2647777234978,
+      "eval_loss": 1.109073281288147,
+      "eval_runtime": 113.3982,
+      "eval_samples_per_second": 256.724,
+      "eval_steps_per_second": 8.025,
+      "step": 116800
+    },
+    {
+      "epoch": 14.46018563751832,
+      "grad_norm": 0.8042516112327576,
+      "learning_rate": 0.00010833703614047328,
+      "loss": 1.2231,
+      "step": 118400
+    },
+    {
+      "epoch": 14.46018563751832,
+      "eval_loss": 1.1034891605377197,
+      "eval_runtime": 113.4506,
+      "eval_samples_per_second": 256.605,
+      "eval_steps_per_second": 8.021,
+      "step": 118400
+    },
+    {
+      "epoch": 14.655593551538837,
+      "grad_norm": 0.8420639038085938,
+      "learning_rate": 0.00010626244325986843,
+      "loss": 1.2205,
+      "step": 120000
+    },
+    {
+      "epoch": 14.655593551538837,
+      "eval_loss": 1.101244330406189,
+      "eval_runtime": 113.3452,
+      "eval_samples_per_second": 256.844,
+      "eval_steps_per_second": 8.029,
+      "step": 120000
+    },
+    {
+      "epoch": 14.851001465559355,
+      "grad_norm": 0.8175898790359497,
+      "learning_rate": 0.00010418514062317943,
+      "loss": 1.2158,
+      "step": 121600
+    },
+    {
+      "epoch": 14.851001465559355,
+      "eval_loss": 1.1000375747680664,
+      "eval_runtime": 113.0595,
+      "eval_samples_per_second": 257.493,
+      "eval_steps_per_second": 8.049,
+      "step": 121600
+    },
+    {
+      "epoch": 15.046409379579872,
+      "grad_norm": 0.8033199906349182,
+      "learning_rate": 0.00010210602707822416,
+      "loss": 1.2126,
+      "step": 123200
+    },
+    {
+      "epoch": 15.046409379579872,
+      "eval_loss": 1.09498131275177,
+      "eval_runtime": 113.2967,
+      "eval_samples_per_second": 256.954,
+      "eval_steps_per_second": 8.032,
+      "step": 123200
+    },
+    {
+      "epoch": 15.241817293600391,
+      "grad_norm": 0.84133380651474,
+      "learning_rate": 0.00010002600225639952,
+      "loss": 1.2065,
+      "step": 124800
+    },
+    {
+      "epoch": 15.241817293600391,
+      "eval_loss": 1.0932313203811646,
+      "eval_runtime": 113.5069,
+      "eval_samples_per_second": 256.478,
+      "eval_steps_per_second": 8.017,
+      "step": 124800
+    },
+    {
+      "epoch": 15.437225207620909,
+      "grad_norm": 0.8350341320037842,
+      "learning_rate": 9.794596618341145e-05,
+      "loss": 1.2052,
+      "step": 126400
+    },
+    {
+      "epoch": 15.437225207620909,
+      "eval_loss": 1.0876601934432983,
+      "eval_runtime": 113.1095,
+      "eval_samples_per_second": 257.379,
+      "eval_steps_per_second": 8.045,
+      "step": 126400
+    },
+    {
+      "epoch": 15.632633121641426,
+      "grad_norm": 0.8618020415306091,
+      "learning_rate": 9.586681888983431e-05,
+      "loss": 1.2018,
+      "step": 128000
+    },
+    {
+      "epoch": 15.632633121641426,
+      "eval_loss": 1.0861952304840088,
+      "eval_runtime": 113.5585,
+      "eval_samples_per_second": 256.361,
+      "eval_steps_per_second": 8.013,
+      "step": 128000
+    },
+    {
+      "epoch": 15.828041035661943,
+      "grad_norm": 0.8563352227210999,
+      "learning_rate": 9.378946002166804e-05,
+      "loss": 1.1988,
+      "step": 129600
+    },
+    {
+      "epoch": 15.828041035661943,
+      "eval_loss": 1.0840609073638916,
+      "eval_runtime": 113.2623,
+      "eval_samples_per_second": 257.032,
+      "eval_steps_per_second": 8.034,
+      "step": 129600
+    },
+    {
+      "epoch": 16.02344894968246,
+      "grad_norm": 0.8290483951568604,
+      "learning_rate": 9.171478845106179e-05,
+      "loss": 1.1932,
+      "step": 131200
+    },
+    {
+      "epoch": 16.02344894968246,
+      "eval_loss": 1.0800191164016724,
+      "eval_runtime": 113.4589,
+      "eval_samples_per_second": 256.586,
+      "eval_steps_per_second": 8.021,
+      "step": 131200
+    },
+    {
+      "epoch": 16.21885686370298,
+      "grad_norm": 0.84539794921875,
+      "learning_rate": 8.964370188737233e-05,
+      "loss": 1.1907,
+      "step": 132800
+    },
+    {
+      "epoch": 16.21885686370298,
+      "eval_loss": 1.0787382125854492,
+      "eval_runtime": 113.3297,
+      "eval_samples_per_second": 256.879,
+      "eval_steps_per_second": 8.03,
+      "step": 132800
+    },
+    {
+      "epoch": 16.4142647777235,
+      "grad_norm": 0.81632000207901,
+      "learning_rate": 8.757709648872583e-05,
+      "loss": 1.1873,
+      "step": 134400
+    },
+    {
+      "epoch": 16.4142647777235,
+      "eval_loss": 1.0750751495361328,
+      "eval_runtime": 113.6133,
+      "eval_samples_per_second": 256.238,
+      "eval_steps_per_second": 8.01,
+      "step": 134400
+    },
+    {
+      "epoch": 16.609672691744017,
+      "grad_norm": 0.8491987586021423,
+      "learning_rate": 8.551586647425051e-05,
+      "loss": 1.1819,
+      "step": 136000
+    },
+    {
+      "epoch": 16.609672691744017,
+      "eval_loss": 1.0718854665756226,
+      "eval_runtime": 113.4371,
+      "eval_samples_per_second": 256.636,
+      "eval_steps_per_second": 8.022,
+      "step": 136000
+    },
+    {
+      "epoch": 16.805080605764534,
+      "grad_norm": 0.8197298645973206,
+      "learning_rate": 8.346090373714858e-05,
+      "loss": 1.1802,
+      "step": 137600
+    },
+    {
+      "epoch": 16.805080605764534,
+      "eval_loss": 1.0670559406280518,
+      "eval_runtime": 113.4213,
+      "eval_samples_per_second": 256.671,
+      "eval_steps_per_second": 8.023,
+      "step": 137600
+    },
+    {
+      "epoch": 17.00048851978505,
+      "grad_norm": 0.8388739824295044,
+      "learning_rate": 8.141309745877437e-05,
+      "loss": 1.1768,
+      "step": 139200
+    },
+    {
+      "epoch": 17.00048851978505,
+      "eval_loss": 1.0644892454147339,
+      "eval_runtime": 112.6464,
+      "eval_samples_per_second": 258.437,
+      "eval_steps_per_second": 8.078,
+      "step": 139200
+    },
+    {
+      "epoch": 17.19589643380557,
+      "grad_norm": 0.9059694409370422,
+      "learning_rate": 7.93733337238861e-05,
+      "loss": 1.1738,
+      "step": 140800
+    },
+    {
+      "epoch": 17.19589643380557,
+      "eval_loss": 1.0664381980895996,
+      "eval_runtime": 112.556,
+      "eval_samples_per_second": 258.645,
+      "eval_steps_per_second": 8.085,
+      "step": 140800
+    },
+    {
+      "epoch": 17.391304347826086,
+      "grad_norm": 0.9374914765357971,
+      "learning_rate": 7.734249513723749e-05,
+      "loss": 1.1699,
+      "step": 142400
+    },
+    {
+      "epoch": 17.391304347826086,
+      "eval_loss": 1.0624644756317139,
+      "eval_runtime": 112.922,
+      "eval_samples_per_second": 257.806,
+      "eval_steps_per_second": 8.059,
+      "step": 142400
+    },
+    {
+      "epoch": 17.586712261846603,
+      "grad_norm": 0.8572972416877747,
+      "learning_rate": 7.532146044167501e-05,
+      "loss": 1.1667,
+      "step": 144000
+    },
+    {
+      "epoch": 17.586712261846603,
+      "eval_loss": 1.0545681715011597,
+      "eval_runtime": 112.6964,
+      "eval_samples_per_second": 258.322,
+      "eval_steps_per_second": 8.075,
+      "step": 144000
+    },
+    {
+      "epoch": 17.78212017586712,
+      "grad_norm": 0.8783763647079468,
+      "learning_rate": 7.33111041379063e-05,
+      "loss": 1.164,
+      "step": 145600
+    },
+    {
+      "epoch": 17.78212017586712,
+      "eval_loss": 1.0523958206176758,
+      "eval_runtime": 112.8245,
+      "eval_samples_per_second": 258.029,
+      "eval_steps_per_second": 8.066,
+      "step": 145600
+    },
+    {
+      "epoch": 17.97752808988764,
+      "grad_norm": 0.8747693300247192,
+      "learning_rate": 7.131229610610423e-05,
+      "loss": 1.1595,
+      "step": 147200
+    },
+    {
+      "epoch": 17.97752808988764,
+      "eval_loss": 1.0538229942321777,
+      "eval_runtime": 112.8739,
+      "eval_samples_per_second": 257.916,
+      "eval_steps_per_second": 8.062,
+      "step": 147200
+    },
+    {
+      "epoch": 18.17293600390816,
+      "grad_norm": 0.8842704892158508,
+      "learning_rate": 6.932590122951006e-05,
+      "loss": 1.1571,
+      "step": 148800
+    },
+    {
+      "epoch": 18.17293600390816,
+      "eval_loss": 1.051466941833496,
+      "eval_runtime": 113.4259,
+      "eval_samples_per_second": 256.661,
+      "eval_steps_per_second": 8.023,
+      "step": 148800
+    },
+    {
+      "epoch": 18.368343917928676,
+      "grad_norm": 0.8712752461433411,
+      "learning_rate": 6.735277902019914e-05,
+      "loss": 1.1552,
+      "step": 150400
+    },
+    {
+      "epoch": 18.368343917928676,
+      "eval_loss": 1.0449974536895752,
+      "eval_runtime": 113.4246,
+      "eval_samples_per_second": 256.664,
+      "eval_steps_per_second": 8.023,
+      "step": 150400
+    },
+    {
+      "epoch": 18.563751831949194,
+      "grad_norm": 0.8981324434280396,
+      "learning_rate": 6.539378324717007e-05,
+      "loss": 1.1503,
+      "step": 152000
+    },
+    {
+      "epoch": 18.563751831949194,
+      "eval_loss": 1.044567584991455,
+      "eval_runtime": 113.3988,
+      "eval_samples_per_second": 256.722,
+      "eval_steps_per_second": 8.025,
+      "step": 152000
+    },
+    {
+      "epoch": 18.75915974596971,
+      "grad_norm": 0.9067941904067993,
+      "learning_rate": 6.344976156691964e-05,
+      "loss": 1.1478,
+      "step": 153600
+    },
+    {
+      "epoch": 18.75915974596971,
+      "eval_loss": 1.045156478881836,
+      "eval_runtime": 112.9533,
+      "eval_samples_per_second": 257.735,
+      "eval_steps_per_second": 8.056,
+      "step": 153600
+    },
+    {
+      "epoch": 18.95456765999023,
+      "grad_norm": 0.8807551264762878,
+      "learning_rate": 6.152155515666206e-05,
+      "loss": 1.145,
+      "step": 155200
+    },
+    {
+      "epoch": 18.95456765999023,
+      "eval_loss": 1.0368510484695435,
+      "eval_runtime": 113.1918,
+      "eval_samples_per_second": 257.192,
+      "eval_steps_per_second": 8.039,
+      "step": 155200
+    },
+    {
+      "epoch": 19.149975574010746,
+      "grad_norm": 0.8794445991516113,
+      "learning_rate": 5.96099983503521e-05,
+      "loss": 1.1403,
+      "step": 156800
+    },
+    {
+      "epoch": 19.149975574010746,
+      "eval_loss": 1.0389199256896973,
+      "eval_runtime": 113.1353,
+      "eval_samples_per_second": 257.32,
+      "eval_steps_per_second": 8.043,
+      "step": 156800
+    },
+    {
+      "epoch": 19.345383488031267,
+      "grad_norm": 0.9533663392066956,
+      "learning_rate": 5.771591827766929e-05,
+      "loss": 1.1403,
+      "step": 158400
+    },
+    {
+      "epoch": 19.345383488031267,
+      "eval_loss": 1.034336805343628,
+      "eval_runtime": 113.0076,
+      "eval_samples_per_second": 257.611,
+      "eval_steps_per_second": 8.053,
+      "step": 158400
+    },
+    {
+      "epoch": 19.540791402051784,
+      "grad_norm": 0.9403147101402283,
+      "learning_rate": 5.5840134506119026e-05,
+      "loss": 1.1355,
+      "step": 160000
+    },
+    {
+      "epoch": 19.540791402051784,
+      "eval_loss": 1.032446026802063,
+      "eval_runtime": 112.7939,
+      "eval_samples_per_second": 258.099,
+      "eval_steps_per_second": 8.068,
+      "step": 160000
+    },
+    {
+      "epoch": 19.7361993160723,
+      "grad_norm": 0.8987004160881042,
+      "learning_rate": 5.398345868640643e-05,
+      "loss": 1.1331,
+      "step": 161600
+    },
+    {
+      "epoch": 19.7361993160723,
+      "eval_loss": 1.0282906293869019,
+      "eval_runtime": 113.1606,
+      "eval_samples_per_second": 257.263,
+      "eval_steps_per_second": 8.042,
+      "step": 161600
+    },
+    {
+      "epoch": 19.93160723009282,
+      "grad_norm": 0.9014713764190674,
+      "learning_rate": 5.2146694201235327e-05,
+      "loss": 1.1295,
+      "step": 163200
+    },
+    {
+      "epoch": 19.93160723009282,
+      "eval_loss": 1.028027057647705,
+      "eval_runtime": 113.1223,
+      "eval_samples_per_second": 257.35,
+      "eval_steps_per_second": 8.044,
+      "step": 163200
+    },
+    {
+      "epoch": 20.127015144113336,
+      "grad_norm": 0.9117501378059387,
+      "learning_rate": 5.033063581768499e-05,
+      "loss": 1.1283,
+      "step": 164800
+    },
+    {
+      "epoch": 20.127015144113336,
+      "eval_loss": 1.0252844095230103,
+      "eval_runtime": 113.2628,
+      "eval_samples_per_second": 257.031,
+      "eval_steps_per_second": 8.034,
+      "step": 164800
+    },
+    {
+      "epoch": 20.322423058133854,
+      "grad_norm": 0.9267609119415283,
+      "learning_rate": 4.8536069343314827e-05,
+      "loss": 1.1249,
+      "step": 166400
+    },
+    {
+      "epoch": 20.322423058133854,
+      "eval_loss": 1.0217244625091553,
+      "eval_runtime": 113.3099,
+      "eval_samples_per_second": 256.924,
+      "eval_steps_per_second": 8.031,
+      "step": 166400
+    },
+    {
+      "epoch": 20.51783097215437,
+      "grad_norm": 0.9269524812698364,
+      "learning_rate": 4.676377128614583e-05,
+      "loss": 1.1209,
+      "step": 168000
+    },
+    {
+      "epoch": 20.51783097215437,
+      "eval_loss": 1.0205986499786377,
+      "eval_runtime": 113.1734,
+      "eval_samples_per_second": 257.234,
+      "eval_steps_per_second": 8.041,
+      "step": 168000
+    },
+    {
+      "epoch": 20.71323888617489,
+      "grad_norm": 0.8992347121238708,
+      "learning_rate": 4.501450851866593e-05,
+      "loss": 1.1199,
+      "step": 169600
+    },
+    {
+      "epoch": 20.71323888617489,
+      "eval_loss": 1.0179848670959473,
+      "eval_runtime": 113.2921,
+      "eval_samples_per_second": 256.964,
+      "eval_steps_per_second": 8.032,
+      "step": 169600
+    },
+    {
+      "epoch": 20.90864680019541,
+      "grad_norm": 0.9611796736717224,
+      "learning_rate": 4.3289037946004674e-05,
+      "loss": 1.1164,
+      "step": 171200
+    },
+    {
+      "epoch": 20.90864680019541,
+      "eval_loss": 1.015058994293213,
+      "eval_runtime": 113.2965,
+      "eval_samples_per_second": 256.954,
+      "eval_steps_per_second": 8.032,
+      "step": 171200
+    },
+    {
+      "epoch": 21.104054714215927,
+      "grad_norm": 0.9261363744735718,
+      "learning_rate": 4.158810617842075e-05,
+      "loss": 1.1125,
+      "step": 172800
+    },
+    {
+      "epoch": 21.104054714215927,
+      "eval_loss": 1.0152846574783325,
+      "eval_runtime": 113.2272,
+      "eval_samples_per_second": 257.111,
+      "eval_steps_per_second": 8.037,
+      "step": 172800
+    },
+    {
+      "epoch": 21.299462628236444,
+      "grad_norm": 0.9231986999511719,
+      "learning_rate": 3.9912449208244075e-05,
+      "loss": 1.1104,
+      "step": 174400
+    },
+    {
+      "epoch": 21.299462628236444,
+      "eval_loss": 1.015738606452942,
+      "eval_runtime": 113.0864,
+      "eval_samples_per_second": 257.431,
+      "eval_steps_per_second": 8.047,
+      "step": 174400
+    },
+    {
+      "epoch": 21.49487054225696,
+      "grad_norm": 0.9297599792480469,
+      "learning_rate": 3.826279209141231e-05,
+      "loss": 1.1087,
+      "step": 176000
+    },
+    {
+      "epoch": 21.49487054225696,
+      "eval_loss": 1.0116798877716064,
+      "eval_runtime": 113.0246,
+      "eval_samples_per_second": 257.572,
+      "eval_steps_per_second": 8.051,
+      "step": 176000
+    },
+    {
+      "epoch": 21.69027845627748,
+      "grad_norm": 0.9401277303695679,
+      "learning_rate": 3.663984863373953e-05,
+      "loss": 1.1052,
+      "step": 177600
+    },
+    {
+      "epoch": 21.69027845627748,
+      "eval_loss": 1.007462978363037,
+      "eval_runtime": 113.0923,
+      "eval_samples_per_second": 257.418,
+      "eval_steps_per_second": 8.047,
+      "step": 177600
+    },
+    {
+      "epoch": 21.885686370297996,
+      "grad_norm": 0.9512894749641418,
+      "learning_rate": 3.504432108205271e-05,
+      "loss": 1.1037,
+      "step": 179200
+    },
+    {
+      "epoch": 21.885686370297996,
+      "eval_loss": 1.0074025392532349,
+      "eval_runtime": 113.1625,
+      "eval_samples_per_second": 257.258,
+      "eval_steps_per_second": 8.042,
+      "step": 179200
+    },
+    {
+      "epoch": 22.081094284318514,
+      "grad_norm": 0.94953453540802,
+      "learning_rate": 3.347689982033e-05,
+      "loss": 1.1,
+      "step": 180800
+    },
+    {
+      "epoch": 22.081094284318514,
+      "eval_loss": 1.0040948390960693,
+      "eval_runtime": 113.5667,
+      "eval_samples_per_second": 256.343,
+      "eval_steps_per_second": 8.013,
+      "step": 180800
+    },
+    {
+      "epoch": 22.27650219833903,
+      "grad_norm": 1.0053319931030273,
+      "learning_rate": 3.193826307097183e-05,
+      "loss": 1.097,
+      "step": 182400
+    },
+    {
+      "epoch": 22.27650219833903,
+      "eval_loss": 1.0029879808425903,
+      "eval_runtime": 113.4145,
+      "eval_samples_per_second": 256.687,
+      "eval_steps_per_second": 8.024,
+      "step": 182400
+    },
+    {
+      "epoch": 22.471910112359552,
+      "grad_norm": 0.9804225564002991,
+      "learning_rate": 3.042907660133447e-05,
+      "loss": 1.0965,
+      "step": 184000
+    },
+    {
+      "epoch": 22.471910112359552,
+      "eval_loss": 1.0013761520385742,
+      "eval_runtime": 113.1629,
+      "eval_samples_per_second": 257.258,
+      "eval_steps_per_second": 8.042,
+      "step": 184000
+    },
+    {
+      "epoch": 22.66731802638007,
+      "grad_norm": 0.9840267896652222,
+      "learning_rate": 2.89499934356528e-05,
+      "loss": 1.0941,
+      "step": 185600
+    },
+    {
+      "epoch": 22.66731802638007,
+      "eval_loss": 1.001320719718933,
+      "eval_runtime": 113.395,
+      "eval_samples_per_second": 256.731,
+      "eval_steps_per_second": 8.025,
+      "step": 185600
+    },
+    {
+      "epoch": 22.862725940400587,
+      "grad_norm": 0.9260748028755188,
+      "learning_rate": 2.7501653572476948e-05,
+      "loss": 1.0902,
+      "step": 187200
+    },
+    {
+      "epoch": 22.862725940400587,
+      "eval_loss": 0.998078465461731,
+      "eval_runtime": 113.1711,
+      "eval_samples_per_second": 257.239,
+      "eval_steps_per_second": 8.041,
+      "step": 187200
+    },
+    {
+      "epoch": 23.058133854421104,
+      "grad_norm": 0.9746040105819702,
+      "learning_rate": 2.6084683707745506e-05,
+      "loss": 1.0887,
+      "step": 188800
+    },
+    {
+      "epoch": 23.058133854421104,
+      "eval_loss": 0.9981781244277954,
+      "eval_runtime": 112.8918,
+      "eval_samples_per_second": 257.875,
+      "eval_steps_per_second": 8.061,
+      "step": 188800
+    },
+    {
+      "epoch": 23.25354176844162,
+      "grad_norm": 0.9964615106582642,
+      "learning_rate": 2.4699696963614248e-05,
+      "loss": 1.087,
+      "step": 190400
+    },
+    {
+      "epoch": 23.25354176844162,
+      "eval_loss": 0.9973635673522949,
+      "eval_runtime": 113.0464,
+      "eval_samples_per_second": 257.523,
+      "eval_steps_per_second": 8.05,
+      "step": 190400
+    },
+    {
+      "epoch": 23.44894968246214,
+      "grad_norm": 0.9735720157623291,
+      "learning_rate": 2.334729262315859e-05,
+      "loss": 1.0847,
+      "step": 192000
+    },
+    {
+      "epoch": 23.44894968246214,
+      "eval_loss": 0.9933942556381226,
+      "eval_runtime": 112.8565,
+      "eval_samples_per_second": 257.956,
+      "eval_steps_per_second": 8.063,
+      "step": 192000
+    },
+    {
+      "epoch": 23.644357596482656,
+      "grad_norm": 1.0612947940826416,
+      "learning_rate": 2.2028055871064014e-05,
+      "loss": 1.0811,
+      "step": 193600
+    },
+    {
+      "epoch": 23.644357596482656,
+      "eval_loss": 0.9925423860549927,
+      "eval_runtime": 112.8304,
+      "eval_samples_per_second": 258.016,
+      "eval_steps_per_second": 8.065,
+      "step": 193600
+    },
+    {
+      "epoch": 23.839765510503174,
+      "grad_norm": 0.9771234393119812,
+      "learning_rate": 2.0742557540417086e-05,
+      "loss": 1.0806,
+      "step": 195200
+    },
+    {
+      "epoch": 23.839765510503174,
+      "eval_loss": 0.9896884560585022,
+      "eval_runtime": 112.7262,
+      "eval_samples_per_second": 258.254,
+      "eval_steps_per_second": 8.073,
+      "step": 195200
+    },
+    {
+      "epoch": 24.035173424523695,
+      "grad_norm": 1.0061612129211426,
+      "learning_rate": 1.9491353865706208e-05,
+      "loss": 1.0787,
+      "step": 196800
+    },
+    {
+      "epoch": 24.035173424523695,
+      "eval_loss": 0.9893516898155212,
+      "eval_runtime": 113.0708,
+      "eval_samples_per_second": 257.467,
+      "eval_steps_per_second": 8.048,
+      "step": 196800
+    },
+    {
+      "epoch": 24.230581338544212,
+      "grad_norm": 0.9902617931365967,
+      "learning_rate": 1.8274986242139203e-05,
+      "loss": 1.0755,
+      "step": 198400
+    },
+    {
+      "epoch": 24.230581338544212,
+      "eval_loss": 0.9926121234893799,
+      "eval_runtime": 113.3137,
+      "eval_samples_per_second": 256.915,
+      "eval_steps_per_second": 8.031,
+      "step": 198400
+    },
+    {
+      "epoch": 24.42598925256473,
+      "grad_norm": 0.9563286900520325,
+      "learning_rate": 1.7093980991381786e-05,
+      "loss": 1.0759,
+      "step": 200000
+    },
+    {
+      "epoch": 24.42598925256473,
+      "eval_loss": 0.9888877272605896,
+      "eval_runtime": 113.167,
+      "eval_samples_per_second": 257.248,
+      "eval_steps_per_second": 8.041,
+      "step": 200000
+    },
+    {
+      "epoch": 24.621397166585247,
+      "grad_norm": 0.9523786902427673,
+      "learning_rate": 1.5948849133818656e-05,
+      "loss": 1.0719,
+      "step": 201600
+    },
+    {
+      "epoch": 24.621397166585247,
+      "eval_loss": 0.9879076480865479,
+      "eval_runtime": 113.1405,
+      "eval_samples_per_second": 257.308,
+      "eval_steps_per_second": 8.043,
+      "step": 201600
+    },
+    {
+      "epoch": 24.816805080605764,
+      "grad_norm": 0.996898889541626,
+      "learning_rate": 1.4840086167435107e-05,
+      "loss": 1.0688,
+      "step": 203200
+    },
+    {
+      "epoch": 24.816805080605764,
+      "eval_loss": 0.9834213852882385,
+      "eval_runtime": 113.246,
+      "eval_samples_per_second": 257.069,
+      "eval_steps_per_second": 8.036,
+      "step": 203200
+    },
+    {
+      "epoch": 25.01221299462628,
+      "grad_norm": 1.096548318862915,
+      "learning_rate": 1.376817185341529e-05,
+      "loss": 1.0676,
+      "step": 204800
+    },
+    {
+      "epoch": 25.01221299462628,
+      "eval_loss": 0.9849600791931152,
+      "eval_runtime": 113.3085,
+      "eval_samples_per_second": 256.927,
+      "eval_steps_per_second": 8.031,
+      "step": 204800
+    },
+    {
+      "epoch": 25.2076209086468,
+      "grad_norm": 1.0107381343841553,
+      "learning_rate": 1.2733570008549767e-05,
+      "loss": 1.067,
+      "step": 206400
+    },
+    {
+      "epoch": 25.2076209086468,
+      "eval_loss": 0.9812195301055908,
+      "eval_runtime": 113.3157,
+      "eval_samples_per_second": 256.911,
+      "eval_steps_per_second": 8.031,
+      "step": 206400
+    },
+    {
+      "epoch": 25.403028822667316,
+      "grad_norm": 1.0007399320602417,
+      "learning_rate": 1.1736728304542287e-05,
+      "loss": 1.0621,
+      "step": 208000
+    },
+    {
+      "epoch": 25.403028822667316,
+      "eval_loss": 0.9833276271820068,
+      "eval_runtime": 113.2967,
+      "eval_samples_per_second": 256.954,
+      "eval_steps_per_second": 8.032,
+      "step": 208000
+    },
+    {
+      "epoch": 25.598436736687837,
+      "grad_norm": 1.0498933792114258,
+      "learning_rate": 1.0778078074302412e-05,
+      "loss": 1.0627,
+      "step": 209600
+    },
+    {
+      "epoch": 25.598436736687837,
+      "eval_loss": 0.9808743596076965,
+      "eval_runtime": 113.2258,
+      "eval_samples_per_second": 257.114,
+      "eval_steps_per_second": 8.037,
+      "step": 209600
+    },
+    {
+      "epoch": 25.793844650708355,
+      "grad_norm": 1.1033276319503784,
+      "learning_rate": 9.85803412530808e-06,
+      "loss": 1.0624,
+      "step": 211200
+    },
+    {
+      "epoch": 25.793844650708355,
+      "eval_loss": 0.9798229336738586,
+      "eval_runtime": 113.2698,
+      "eval_samples_per_second": 257.015,
+      "eval_steps_per_second": 8.034,
+      "step": 211200
+    },
+    {
+      "epoch": 25.989252564728872,
+      "grad_norm": 1.0208085775375366,
+      "learning_rate": 8.976994560118401e-06,
+      "loss": 1.0614,
+      "step": 212800
+    },
+    {
+      "epoch": 25.989252564728872,
+      "eval_loss": 0.9794700145721436,
+      "eval_runtime": 113.4354,
+      "eval_samples_per_second": 256.64,
+      "eval_steps_per_second": 8.022,
+      "step": 212800
+    },
+    {
+      "epoch": 26.18466047874939,
+      "grad_norm": 1.050437092781067,
+      "learning_rate": 8.135340604115083e-06,
+      "loss": 1.0602,
+      "step": 214400
+    },
+    {
+      "epoch": 26.18466047874939,
+      "eval_loss": 0.9784888029098511,
+      "eval_runtime": 113.3908,
+      "eval_samples_per_second": 256.74,
+      "eval_steps_per_second": 8.025,
+      "step": 214400
+    },
+    {
+      "epoch": 26.380068392769907,
+      "grad_norm": 1.0029643774032593,
+      "learning_rate": 7.333436440546337e-06,
+      "loss": 1.0551,
+      "step": 216000
+    },
+    {
+      "epoch": 26.380068392769907,
+      "eval_loss": 0.978079617023468,
+      "eval_runtime": 113.3592,
+      "eval_samples_per_second": 256.812,
+      "eval_steps_per_second": 8.028,
+      "step": 216000
+    },
+    {
+      "epoch": 26.575476306790424,
+      "grad_norm": 0.9882290959358215,
+      "learning_rate": 6.571629052944928e-06,
+      "loss": 1.0554,
+      "step": 217600
+    },
+    {
+      "epoch": 26.575476306790424,
+      "eval_loss": 0.9771565794944763,
+      "eval_runtime": 113.2886,
+      "eval_samples_per_second": 256.972,
+      "eval_steps_per_second": 8.033,
+      "step": 217600
+    },
+    {
+      "epoch": 26.77088422081094,
+      "grad_norm": 1.0697990655899048,
+      "learning_rate": 5.850248074988618e-06,
+      "loss": 1.0539,
+      "step": 219200
+    },
+    {
+      "epoch": 26.77088422081094,
+      "eval_loss": 0.9780036211013794,
+      "eval_runtime": 113.0821,
+      "eval_samples_per_second": 257.441,
+      "eval_steps_per_second": 8.047,
+      "step": 219200
+    },
+    {
+      "epoch": 26.96629213483146,
+      "grad_norm": 1.0118811130523682,
+      "learning_rate": 5.169605647867792e-06,
+      "loss": 1.0533,
+      "step": 220800
+    },
+    {
+      "epoch": 26.96629213483146,
+      "eval_loss": 0.9773528575897217,
+      "eval_runtime": 113.4381,
+      "eval_samples_per_second": 256.633,
+      "eval_steps_per_second": 8.022,
+      "step": 220800
+    },
+    {
+      "epoch": 27.16170004885198,
+      "grad_norm": 1.0081678628921509,
+      "learning_rate": 4.5299962852221935e-06,
+      "loss": 1.0505,
+      "step": 222400
+    },
+    {
+      "epoch": 27.16170004885198,
+      "eval_loss": 0.976798951625824,
+      "eval_runtime": 113.7993,
+      "eval_samples_per_second": 255.819,
+      "eval_steps_per_second": 7.997,
+      "step": 222400
+    },
+    {
+      "epoch": 27.357107962872497,
+      "grad_norm": 1.0098705291748047,
+      "learning_rate": 3.931696745704927e-06,
+      "loss": 1.0493,
+      "step": 224000
+    },
+    {
+      "epoch": 27.357107962872497,
+      "eval_loss": 0.9746509790420532,
+      "eval_runtime": 113.5227,
+      "eval_samples_per_second": 256.442,
+      "eval_steps_per_second": 8.016,
+      "step": 224000
+    },
+    {
+      "epoch": 27.552515876893015,
+      "grad_norm": 0.9799443483352661,
+      "learning_rate": 3.374965913229211e-06,
+      "loss": 1.0499,
+      "step": 225600
+    },
+    {
+      "epoch": 27.552515876893015,
+      "eval_loss": 0.9754048585891724,
+      "eval_runtime": 113.9041,
+      "eval_samples_per_second": 255.583,
+      "eval_steps_per_second": 7.989,
+      "step": 225600
+    },
+    {
+      "epoch": 27.747923790913532,
+      "grad_norm": 1.1257084608078003,
+      "learning_rate": 2.8600446849493812e-06,
+      "loss": 1.0466,
+      "step": 227200
+    },
+    {
+      "epoch": 27.747923790913532,
+      "eval_loss": 0.9729560017585754,
+      "eval_runtime": 114.0446,
+      "eval_samples_per_second": 255.269,
+      "eval_steps_per_second": 7.979,
+      "step": 227200
+    },
+    {
+      "epoch": 27.94333170493405,
+      "grad_norm": 1.0175942182540894,
+      "learning_rate": 2.3871558670248374e-06,
+      "loss": 1.0451,
+      "step": 228800
+    },
+    {
+      "epoch": 27.94333170493405,
+      "eval_loss": 0.9726097583770752,
+      "eval_runtime": 113.8136,
+      "eval_samples_per_second": 255.787,
+      "eval_steps_per_second": 7.996,
+      "step": 228800
+    },
+    {
+      "epoch": 28.138739618954567,
+      "grad_norm": 1.0665676593780518,
+      "learning_rate": 1.9565040782119183e-06,
+      "loss": 1.046,
+      "step": 230400
+    },
+    {
+      "epoch": 28.138739618954567,
+      "eval_loss": 0.9738107919692993,
+      "eval_runtime": 113.6919,
+      "eval_samples_per_second": 256.06,
+      "eval_steps_per_second": 8.004,
+      "step": 230400
+    },
+    {
+      "epoch": 28.334147532975084,
+      "grad_norm": 0.9979843497276306,
+      "learning_rate": 1.5682756613254578e-06,
+      "loss": 1.0461,
+      "step": 232000
+    },
+    {
+      "epoch": 28.334147532975084,
+      "eval_loss": 0.9727683663368225,
+      "eval_runtime": 113.5319,
+      "eval_samples_per_second": 256.421,
+      "eval_steps_per_second": 8.015,
+      "step": 232000
+    },
+    {
+      "epoch": 28.5295554469956,
+      "grad_norm": 1.0315582752227783,
+      "learning_rate": 1.2226386026083835e-06,
+      "loss": 1.044,
+      "step": 233600
+    },
+    {
+      "epoch": 28.5295554469956,
+      "eval_loss": 0.9729660749435425,
+      "eval_runtime": 113.6603,
+      "eval_samples_per_second": 256.132,
+      "eval_steps_per_second": 8.006,
+      "step": 233600
+    },
+    {
+      "epoch": 28.724963361016123,
+      "grad_norm": 1.0685278177261353,
+      "learning_rate": 9.19742459044104e-07,
+      "loss": 1.0445,
+      "step": 235200
+    },
+    {
+      "epoch": 28.724963361016123,
+      "eval_loss": 0.9717814326286316,
+      "eval_runtime": 113.7616,
+      "eval_samples_per_second": 255.904,
+      "eval_steps_per_second": 7.999,
+      "step": 235200
+    },
+    {
+      "epoch": 28.92037127503664,
+      "grad_norm": 1.077459454536438,
+      "learning_rate": 6.597182936433189e-07,
+      "loss": 1.0412,
+      "step": 236800
+    },
+    {
+      "epoch": 28.92037127503664,
+      "eval_loss": 0.9729886054992676,
+      "eval_runtime": 113.5403,
+      "eval_samples_per_second": 256.402,
+      "eval_steps_per_second": 8.015,
+      "step": 236800
+    },
+    {
+      "epoch": 29.115779189057157,
+      "grad_norm": 1.0783263444900513,
+      "learning_rate": 4.426786187330612e-07,
+      "loss": 1.042,
+      "step": 238400
+    },
+    {
+      "epoch": 29.115779189057157,
+      "eval_loss": 0.9737541675567627,
+      "eval_runtime": 113.3216,
+      "eval_samples_per_second": 256.897,
+      "eval_steps_per_second": 8.03,
+      "step": 238400
+    },
+    {
+      "epoch": 29.311187103077675,
+      "grad_norm": 1.0996705293655396,
+      "learning_rate": 2.6871734727274e-07,
+      "loss": 1.04,
+      "step": 240000
+    },
+    {
+      "epoch": 29.311187103077675,
+      "eval_loss": 0.9717140197753906,
+      "eval_runtime": 113.3888,
+      "eval_samples_per_second": 256.745,
+      "eval_steps_per_second": 8.025,
+      "step": 240000
+    },
+    {
+      "epoch": 29.506595017098192,
+      "grad_norm": 1.12758207321167,
+      "learning_rate": 1.3790975221799062e-07,
+      "loss": 1.0402,
+      "step": 241600
+    },
+    {
+      "epoch": 29.506595017098192,
+      "eval_loss": 0.9729249477386475,
+      "eval_runtime": 113.1761,
+      "eval_samples_per_second": 257.227,
+      "eval_steps_per_second": 8.041,
+      "step": 241600
+    },
+    {
+      "epoch": 29.70200293111871,
+      "grad_norm": 1.1545348167419434,
+      "learning_rate": 5.03124339501504e-08,
+      "loss": 1.0398,
+      "step": 243200
+    },
+    {
+      "epoch": 29.70200293111871,
+      "eval_loss": 0.9734637141227722,
+      "eval_runtime": 113.3684,
+      "eval_samples_per_second": 256.791,
+      "eval_steps_per_second": 8.027,
+      "step": 243200
+    },
+    {
+      "epoch": 29.897410845139227,
+      "grad_norm": 1.0583094358444214,
+      "learning_rate": 5.963295785271772e-09,
+      "loss": 1.038,
+      "step": 244800
+    },
+    {
+      "epoch": 29.897410845139227,
+      "eval_loss": 0.9716248512268066,
+      "eval_runtime": 113.626,
+      "eval_samples_per_second": 256.209,
+      "eval_steps_per_second": 8.009,
+      "step": 244800
+    }
+  ],
+  "logging_steps": 1600,
+  "max_steps": 245640,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 30,
+  "save_steps": 1600,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.783639331402416e+18,
+  "train_batch_size": 32,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bae1a3a3a4418bed79fe9d41130695ea291aafba55a576bef30dfb0abf4a1e2f
+size 5368

vocab.json

@@ -0,0 +1,40 @@
+{
+  "[PAD]": 0,
+  "[MASK]": 1,
+  "[CLS]": 2,
+  "[SEP]": 3,
+  "[UNK]": 4,
+  "\n": 5,
+  " ": 6,
+  "'": 7,
+  "-": 8,
+  ".": 9,
+  "α": 10,
+  "β": 11,
+  "γ": 12,
+  "δ": 13,
+  "ε": 14,
+  "ζ": 15,
+  "η": 16,
+  "θ": 17,
+  "ι": 18,
+  "κ": 19,
+  "λ": 20,
+  "μ": 21,
+  "ν": 22,
+  "ξ": 23,
+  "ο": 24,
+  "π": 25,
+  "ρ": 26,
+  "ς": 27,
+  "σ": 28,
+  "τ": 29,
+  "υ": 30,
+  "φ": 31,
+  "χ": 32,
+  "ψ": 33,
+  "ω": 34,
+  "ϛ": 35,
+  "ϼ": 36,
+  "☧": 37
+}