Upload checkpoint-6231

.gitattributes

@@ -35,3 +35,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 checkpoint-2077/tokenizer.json filter=lfs diff=lfs merge=lfs -text
 checkpoint-4154/tokenizer.json filter=lfs diff=lfs merge=lfs -text
+checkpoint-6231/tokenizer.json filter=lfs diff=lfs merge=lfs -text

checkpoint-6231/chat_template.jinja

@@ -0,0 +1,4 @@
+{% if messages[0]['role'] == 'system' %}{{ raise_exception('System role not supported') }}{% endif %}{% for message in 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 + '
+' + message['content'] | trim + '<end_of_turn>
+' }}{% endfor %}{% if add_generation_prompt %}{{'<start_of_turn>model
+'}}{% endif %}

checkpoint-6231/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:32d402ea734649aa80ada025f3fcea6dc90104a395a4cd92d82ea07b245108f0
+size 37657227

checkpoint-6231/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db2fa52c242662db8722e857af5dfd0850cbf376a4270adec6b01da07521ebf5
+size 15429

checkpoint-6231/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e96957fe68dd4a282692d04233291b51b46a970c05e566d98103f6a7d959500
+size 15429

checkpoint-6231/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5224f334e4b90bdf0d7916a84882c2603ea399946407f4b0acc0694ef7a369d9
+size 15429

checkpoint-6231/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ca39909e1d873332ebfe2a478093e1b1226892ebc7e07d9c8427bc08dca0afc
+size 15429

checkpoint-6231/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:00dffc00e29cd34a07a7d050be731e4805ec127bc82abd0466a19d71fe5e79d3
+size 1465

checkpoint-6231/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bdc0fd8410449f9ab91a0bc149a04e4f2565ef5d608b588c971e500863286adf
+size 34362526

checkpoint-6231/tokenizer_config.json

@@ -0,0 +1,17 @@
+{
+  "backend": "tokenizers",
+  "bos_token": "<bos>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<eos>",
+  "is_local": false,
+  "local_files_only": false,
+  "mask_token": "<mask>",
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "<pad>",
+  "padding_side": "right",
+  "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
+  "tokenizer_class": "GemmaTokenizer",
+  "unk_token": "<unk>",
+  "use_default_system_prompt": false
+}

checkpoint-6231/trainable_params.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:896cdd83dd55efe33b8492fef1cb673169a95bb122011607be7354190650ee99
+size 18815091

checkpoint-6231/trainer_state.json

@@ -0,0 +1,1808 @@
+{
+  "best_global_step": 6231,
+  "best_metric": 2.7634668350219727,
+  "best_model_checkpoint": "outputs/chess-explainer/checkpoint-6231",
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 6231,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00048146364949446316,
+      "grad_norm": 650.5328369140625,
+      "learning_rate": 0.0,
+      "loss": 18.217763900756836,
+      "step": 1
+    },
+    {
+      "epoch": 0.01203659123736158,
+      "grad_norm": 88.18751525878906,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 14.24288813273112,
+      "step": 25
+    },
+    {
+      "epoch": 0.02407318247472316,
+      "grad_norm": 22.323110580444336,
+      "learning_rate": 1.5705128205128205e-05,
+      "loss": 7.188988037109375,
+      "step": 50
+    },
+    {
+      "epoch": 0.036109773712084736,
+      "grad_norm": 14.299266815185547,
+      "learning_rate": 2.3717948717948718e-05,
+      "loss": 4.387466735839844,
+      "step": 75
+    },
+    {
+      "epoch": 0.04814636494944632,
+      "grad_norm": 13.159931182861328,
+      "learning_rate": 3.1730769230769234e-05,
+      "loss": 3.8534796142578127,
+      "step": 100
+    },
+    {
+      "epoch": 0.0601829561868079,
+      "grad_norm": 16.505720138549805,
+      "learning_rate": 3.974358974358974e-05,
+      "loss": 3.6811083984375,
+      "step": 125
+    },
+    {
+      "epoch": 0.07221954742416947,
+      "grad_norm": 14.32520866394043,
+      "learning_rate": 4.775641025641026e-05,
+      "loss": 3.5446063232421876,
+      "step": 150
+    },
+    {
+      "epoch": 0.08425613866153106,
+      "grad_norm": 15.38774299621582,
+      "learning_rate": 5.576923076923077e-05,
+      "loss": 3.47775390625,
+      "step": 175
+    },
+    {
+      "epoch": 0.09629272989889263,
+      "grad_norm": 11.126094818115234,
+      "learning_rate": 6.378205128205128e-05,
+      "loss": 3.4696426391601562,
+      "step": 200
+    },
+    {
+      "epoch": 0.10832932113625421,
+      "grad_norm": 10.559870719909668,
+      "learning_rate": 7.17948717948718e-05,
+      "loss": 3.4579669189453126,
+      "step": 225
+    },
+    {
+      "epoch": 0.1203659123736158,
+      "grad_norm": 11.133081436157227,
+      "learning_rate": 7.980769230769231e-05,
+      "loss": 3.359324951171875,
+      "step": 250
+    },
+    {
+      "epoch": 0.13240250361097738,
+      "grad_norm": 10.544161796569824,
+      "learning_rate": 8.782051282051283e-05,
+      "loss": 3.298221435546875,
+      "step": 275
+    },
+    {
+      "epoch": 0.14443909484833894,
+      "grad_norm": 8.630677223205566,
+      "learning_rate": 9.583333333333334e-05,
+      "loss": 3.361368408203125,
+      "step": 300
+    },
+    {
+      "epoch": 0.15647568608570053,
+      "grad_norm": 8.548908233642578,
+      "learning_rate": 0.00010384615384615386,
+      "loss": 3.345328674316406,
+      "step": 325
+    },
+    {
+      "epoch": 0.16851227732306212,
+      "grad_norm": 8.20425796508789,
+      "learning_rate": 0.00011185897435897436,
+      "loss": 3.280472717285156,
+      "step": 350
+    },
+    {
+      "epoch": 0.18054886856042368,
+      "grad_norm": 9.912631034851074,
+      "learning_rate": 0.00011987179487179487,
+      "loss": 3.3183026123046875,
+      "step": 375
+    },
+    {
+      "epoch": 0.19258545979778527,
+      "grad_norm": 9.256693840026855,
+      "learning_rate": 0.00012788461538461537,
+      "loss": 3.2713458251953127,
+      "step": 400
+    },
+    {
+      "epoch": 0.20462205103514686,
+      "grad_norm": 8.893656730651855,
+      "learning_rate": 0.0001358974358974359,
+      "loss": 3.2640109252929688,
+      "step": 425
+    },
+    {
+      "epoch": 0.21665864227250842,
+      "grad_norm": 8.918048858642578,
+      "learning_rate": 0.0001439102564102564,
+      "loss": 3.2969732666015625,
+      "step": 450
+    },
+    {
+      "epoch": 0.22869523350987,
+      "grad_norm": 8.13305377960205,
+      "learning_rate": 0.00015192307692307692,
+      "loss": 3.25054443359375,
+      "step": 475
+    },
+    {
+      "epoch": 0.2407318247472316,
+      "grad_norm": 7.971743583679199,
+      "learning_rate": 0.00015993589743589745,
+      "loss": 3.234297790527344,
+      "step": 500
+    },
+    {
+      "epoch": 0.25276841598459315,
+      "grad_norm": 7.267224311828613,
+      "learning_rate": 0.00016794871794871796,
+      "loss": 3.2479425048828126,
+      "step": 525
+    },
+    {
+      "epoch": 0.26480500722195477,
+      "grad_norm": 8.255003929138184,
+      "learning_rate": 0.00017596153846153846,
+      "loss": 3.2425189208984375,
+      "step": 550
+    },
+    {
+      "epoch": 0.27684159845931633,
+      "grad_norm": 7.8051042556762695,
+      "learning_rate": 0.00018397435897435897,
+      "loss": 3.2422137451171875,
+      "step": 575
+    },
+    {
+      "epoch": 0.2888781896966779,
+      "grad_norm": 8.21413516998291,
+      "learning_rate": 0.0001919871794871795,
+      "loss": 3.2551699829101564,
+      "step": 600
+    },
+    {
+      "epoch": 0.3009147809340395,
+      "grad_norm": 7.534178733825684,
+      "learning_rate": 0.0002,
+      "loss": 3.2088958740234377,
+      "step": 625
+    },
+    {
+      "epoch": 0.31295137217140107,
+      "grad_norm": 7.962012767791748,
+      "learning_rate": 0.00019999924007355532,
+      "loss": 3.2368560791015626,
+      "step": 650
+    },
+    {
+      "epoch": 0.3249879634087626,
+      "grad_norm": 7.770707130432129,
+      "learning_rate": 0.00019999696030577107,
+      "loss": 3.2101596069335936,
+      "step": 675
+    },
+    {
+      "epoch": 0.33702455464612424,
+      "grad_norm": 7.733644008636475,
+      "learning_rate": 0.00019999316073129634,
+      "loss": 3.2300433349609374,
+      "step": 700
+    },
+    {
+      "epoch": 0.3490611458834858,
+      "grad_norm": 7.780461311340332,
+      "learning_rate": 0.00019998784140787908,
+      "loss": 3.2330572509765627,
+      "step": 725
+    },
+    {
+      "epoch": 0.36109773712084736,
+      "grad_norm": 6.770570278167725,
+      "learning_rate": 0.00019998100241636516,
+      "loss": 3.2057620239257814,
+      "step": 750
+    },
+    {
+      "epoch": 0.373134328358209,
+      "grad_norm": 5.945966720581055,
+      "learning_rate": 0.0001999726438606972,
+      "loss": 3.2183990478515625,
+      "step": 775
+    },
+    {
+      "epoch": 0.38517091959557054,
+      "grad_norm": 7.591160774230957,
+      "learning_rate": 0.000199962765867913,
+      "loss": 3.191751708984375,
+      "step": 800
+    },
+    {
+      "epoch": 0.3972075108329321,
+      "grad_norm": 7.020476341247559,
+      "learning_rate": 0.00019995136858814346,
+      "loss": 3.181463623046875,
+      "step": 825
+    },
+    {
+      "epoch": 0.4092441020702937,
+      "grad_norm": 7.284232139587402,
+      "learning_rate": 0.00019993845219461046,
+      "loss": 3.1981622314453126,
+      "step": 850
+    },
+    {
+      "epoch": 0.4212806933076553,
+      "grad_norm": 6.910461902618408,
+      "learning_rate": 0.00019992401688362422,
+      "loss": 3.14352294921875,
+      "step": 875
+    },
+    {
+      "epoch": 0.43331728454501683,
+      "grad_norm": 6.141090393066406,
+      "learning_rate": 0.00019990806287458024,
+      "loss": 3.1020236206054688,
+      "step": 900
+    },
+    {
+      "epoch": 0.44535387578237845,
+      "grad_norm": 6.360198497772217,
+      "learning_rate": 0.00019989059040995594,
+      "loss": 3.233653564453125,
+      "step": 925
+    },
+    {
+      "epoch": 0.45739046701974,
+      "grad_norm": 6.173579692840576,
+      "learning_rate": 0.00019987159975530712,
+      "loss": 3.1653778076171877,
+      "step": 950
+    },
+    {
+      "epoch": 0.46942705825710157,
+      "grad_norm": 6.959299087524414,
+      "learning_rate": 0.00019985109119926377,
+      "loss": 3.1667141723632812,
+      "step": 975
+    },
+    {
+      "epoch": 0.4814636494944632,
+      "grad_norm": 6.9440107345581055,
+      "learning_rate": 0.00019982906505352577,
+      "loss": 3.1662142944335936,
+      "step": 1000
+    },
+    {
+      "epoch": 0.49350024073182475,
+      "grad_norm": 7.002237319946289,
+      "learning_rate": 0.00019980552165285815,
+      "loss": 3.1565814208984375,
+      "step": 1025
+    },
+    {
+      "epoch": 0.5055368319691863,
+      "grad_norm": 6.859927177429199,
+      "learning_rate": 0.00019978046135508596,
+      "loss": 3.177081298828125,
+      "step": 1050
+    },
+    {
+      "epoch": 0.5175734232065479,
+      "grad_norm": 6.390033721923828,
+      "learning_rate": 0.00019975388454108884,
+      "loss": 3.1408489990234374,
+      "step": 1075
+    },
+    {
+      "epoch": 0.5296100144439095,
+      "grad_norm": 6.612856864929199,
+      "learning_rate": 0.00019972579161479524,
+      "loss": 3.125029296875,
+      "step": 1100
+    },
+    {
+      "epoch": 0.5416466056812711,
+      "grad_norm": 6.74856424331665,
+      "learning_rate": 0.0001996961830031764,
+      "loss": 3.1270068359375,
+      "step": 1125
+    },
+    {
+      "epoch": 0.5536831969186327,
+      "grad_norm": 5.670539379119873,
+      "learning_rate": 0.00019966505915623963,
+      "loss": 3.079051208496094,
+      "step": 1150
+    },
+    {
+      "epoch": 0.5657197881559942,
+      "grad_norm": 6.464207649230957,
+      "learning_rate": 0.00019963242054702157,
+      "loss": 3.1350054931640625,
+      "step": 1175
+    },
+    {
+      "epoch": 0.5777563793933558,
+      "grad_norm": 5.980476379394531,
+      "learning_rate": 0.0001995982676715811,
+      "loss": 3.1213824462890627,
+      "step": 1200
+    },
+    {
+      "epoch": 0.5897929706307173,
+      "grad_norm": 6.767940044403076,
+      "learning_rate": 0.00019956260104899166,
+      "loss": 3.107749328613281,
+      "step": 1225
+    },
+    {
+      "epoch": 0.601829561868079,
+      "grad_norm": 6.736208915710449,
+      "learning_rate": 0.00019952542122133346,
+      "loss": 3.1330535888671873,
+      "step": 1250
+    },
+    {
+      "epoch": 0.6138661531054406,
+      "grad_norm": 6.237167835235596,
+      "learning_rate": 0.00019948672875368524,
+      "loss": 3.090823974609375,
+      "step": 1275
+    },
+    {
+      "epoch": 0.6259027443428021,
+      "grad_norm": 5.9626288414001465,
+      "learning_rate": 0.00019944652423411547,
+      "loss": 3.068702392578125,
+      "step": 1300
+    },
+    {
+      "epoch": 0.6379393355801637,
+      "grad_norm": 6.667422294616699,
+      "learning_rate": 0.00019940480827367377,
+      "loss": 3.1129641723632813,
+      "step": 1325
+    },
+    {
+      "epoch": 0.6499759268175253,
+      "grad_norm": 6.018889427185059,
+      "learning_rate": 0.0001993615815063814,
+      "loss": 3.0886264038085938,
+      "step": 1350
+    },
+    {
+      "epoch": 0.6620125180548868,
+      "grad_norm": 5.504490375518799,
+      "learning_rate": 0.00019931684458922159,
+      "loss": 3.0925244140625,
+      "step": 1375
+    },
+    {
+      "epoch": 0.6740491092922485,
+      "grad_norm": 6.392239093780518,
+      "learning_rate": 0.00019927059820212969,
+      "loss": 3.067706298828125,
+      "step": 1400
+    },
+    {
+      "epoch": 0.68608570052961,
+      "grad_norm": 5.7260637283325195,
+      "learning_rate": 0.0001992228430479827,
+      "loss": 3.1082818603515623,
+      "step": 1425
+    },
+    {
+      "epoch": 0.6981222917669716,
+      "grad_norm": 5.574620723724365,
+      "learning_rate": 0.00019917357985258877,
+      "loss": 3.0852099609375,
+      "step": 1450
+    },
+    {
+      "epoch": 0.7101588830043332,
+      "grad_norm": 5.973651885986328,
+      "learning_rate": 0.00019912280936467598,
+      "loss": 3.032557373046875,
+      "step": 1475
+    },
+    {
+      "epoch": 0.7221954742416947,
+      "grad_norm": 6.408794403076172,
+      "learning_rate": 0.00019907053235588103,
+      "loss": 3.0586279296875,
+      "step": 1500
+    },
+    {
+      "epoch": 0.7342320654790563,
+      "grad_norm": 5.924404144287109,
+      "learning_rate": 0.00019901674962073757,
+      "loss": 3.0331820678710937,
+      "step": 1525
+    },
+    {
+      "epoch": 0.746268656716418,
+      "grad_norm": 6.371328830718994,
+      "learning_rate": 0.00019896146197666408,
+      "loss": 3.044822998046875,
+      "step": 1550
+    },
+    {
+      "epoch": 0.7583052479537795,
+      "grad_norm": 5.798267841339111,
+      "learning_rate": 0.00019890467026395137,
+      "loss": 3.086248779296875,
+      "step": 1575
+    },
+    {
+      "epoch": 0.7703418391911411,
+      "grad_norm": 6.200769424438477,
+      "learning_rate": 0.00019884637534574995,
+      "loss": 3.0701373291015623,
+      "step": 1600
+    },
+    {
+      "epoch": 0.7823784304285026,
+      "grad_norm": 5.729808807373047,
+      "learning_rate": 0.00019878657810805682,
+      "loss": 3.033974914550781,
+      "step": 1625
+    },
+    {
+      "epoch": 0.7944150216658642,
+      "grad_norm": 5.3061699867248535,
+      "learning_rate": 0.00019872527945970206,
+      "loss": 3.023597412109375,
+      "step": 1650
+    },
+    {
+      "epoch": 0.8064516129032258,
+      "grad_norm": 6.059074878692627,
+      "learning_rate": 0.00019866248033233486,
+      "loss": 3.038387451171875,
+      "step": 1675
+    },
+    {
+      "epoch": 0.8184882041405874,
+      "grad_norm": 5.904861927032471,
+      "learning_rate": 0.00019859818168040962,
+      "loss": 3.0605734252929686,
+      "step": 1700
+    },
+    {
+      "epoch": 0.830524795377949,
+      "grad_norm": 7.235841751098633,
+      "learning_rate": 0.00019853238448117128,
+      "loss": 3.0388153076171873,
+      "step": 1725
+    },
+    {
+      "epoch": 0.8425613866153105,
+      "grad_norm": 5.9548869132995605,
+      "learning_rate": 0.00019846508973464047,
+      "loss": 3.052736511230469,
+      "step": 1750
+    },
+    {
+      "epoch": 0.8545979778526721,
+      "grad_norm": 6.491075038909912,
+      "learning_rate": 0.00019839629846359828,
+      "loss": 3.04111572265625,
+      "step": 1775
+    },
+    {
+      "epoch": 0.8666345690900337,
+      "grad_norm": 6.274312496185303,
+      "learning_rate": 0.0001983260117135709,
+      "loss": 3.0903143310546874,
+      "step": 1800
+    },
+    {
+      "epoch": 0.8786711603273952,
+      "grad_norm": 6.714303016662598,
+      "learning_rate": 0.0001982542305528135,
+      "loss": 3.0477191162109376,
+      "step": 1825
+    },
+    {
+      "epoch": 0.8907077515647569,
+      "grad_norm": 6.037519931793213,
+      "learning_rate": 0.00019818095607229412,
+      "loss": 3.0680242919921876,
+      "step": 1850
+    },
+    {
+      "epoch": 0.9027443428021185,
+      "grad_norm": 6.358810901641846,
+      "learning_rate": 0.00019810618938567712,
+      "loss": 3.0358065795898437,
+      "step": 1875
+    },
+    {
+      "epoch": 0.91478093403948,
+      "grad_norm": 5.641435623168945,
+      "learning_rate": 0.00019802993162930607,
+      "loss": 3.0244366455078127,
+      "step": 1900
+    },
+    {
+      "epoch": 0.9268175252768416,
+      "grad_norm": 6.10072135925293,
+      "learning_rate": 0.00019795218396218673,
+      "loss": 3.0294232177734375,
+      "step": 1925
+    },
+    {
+      "epoch": 0.9388541165142031,
+      "grad_norm": 5.697592735290527,
+      "learning_rate": 0.00019787294756596927,
+      "loss": 2.9873196411132814,
+      "step": 1950
+    },
+    {
+      "epoch": 0.9508907077515648,
+      "grad_norm": 5.733670711517334,
+      "learning_rate": 0.0001977922236449303,
+      "loss": 3.004986572265625,
+      "step": 1975
+    },
+    {
+      "epoch": 0.9629272989889264,
+      "grad_norm": 6.309650897979736,
+      "learning_rate": 0.00019771001342595471,
+      "loss": 3.00642578125,
+      "step": 2000
+    },
+    {
+      "epoch": 0.9749638902262879,
+      "grad_norm": 6.235594749450684,
+      "learning_rate": 0.0001976263181585169,
+      "loss": 2.9961761474609374,
+      "step": 2025
+    },
+    {
+      "epoch": 0.9870004814636495,
+      "grad_norm": 6.950772285461426,
+      "learning_rate": 0.00019754113911466176,
+      "loss": 3.034189453125,
+      "step": 2050
+    },
+    {
+      "epoch": 0.999037072701011,
+      "grad_norm": 5.884391784667969,
+      "learning_rate": 0.0001974544775889855,
+      "loss": 3.075783996582031,
+      "step": 2075
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 2.9775943756103516,
+      "eval_runtime": 7.6569,
+      "eval_samples_per_second": 195.902,
+      "eval_steps_per_second": 6.138,
+      "step": 2077
+    },
+    {
+      "epoch": 1.0110736639383726,
+      "grad_norm": 5.676021575927734,
+      "learning_rate": 0.0001973663348986158,
+      "loss": 2.86690673828125,
+      "step": 2100
+    },
+    {
+      "epoch": 1.0231102551757343,
+      "grad_norm": 6.018618106842041,
+      "learning_rate": 0.00019727671238319188,
+      "loss": 2.8802978515625,
+      "step": 2125
+    },
+    {
+      "epoch": 1.0351468464130957,
+      "grad_norm": 6.222283840179443,
+      "learning_rate": 0.00019718561140484413,
+      "loss": 2.8613705444335937,
+      "step": 2150
+    },
+    {
+      "epoch": 1.0471834376504574,
+      "grad_norm": 5.6597208976745605,
+      "learning_rate": 0.0001970930333481734,
+      "loss": 2.856763916015625,
+      "step": 2175
+    },
+    {
+      "epoch": 1.059220028887819,
+      "grad_norm": 6.7447662353515625,
+      "learning_rate": 0.00019699897962022997,
+      "loss": 2.9224319458007812,
+      "step": 2200
+    },
+    {
+      "epoch": 1.0712566201251805,
+      "grad_norm": 7.006275653839111,
+      "learning_rate": 0.0001969034516504922,
+      "loss": 2.93245361328125,
+      "step": 2225
+    },
+    {
+      "epoch": 1.0832932113625422,
+      "grad_norm": 5.889551162719727,
+      "learning_rate": 0.00019680645089084454,
+      "loss": 2.875589599609375,
+      "step": 2250
+    },
+    {
+      "epoch": 1.0953298025999036,
+      "grad_norm": 6.063162326812744,
+      "learning_rate": 0.000196707978815556,
+      "loss": 2.874587097167969,
+      "step": 2275
+    },
+    {
+      "epoch": 1.1073663938372653,
+      "grad_norm": 6.2372145652771,
+      "learning_rate": 0.00019660803692125715,
+      "loss": 2.8780206298828124,
+      "step": 2300
+    },
+    {
+      "epoch": 1.1194029850746268,
+      "grad_norm": 6.315883159637451,
+      "learning_rate": 0.00019650662672691783,
+      "loss": 2.8887982177734375,
+      "step": 2325
+    },
+    {
+      "epoch": 1.1314395763119884,
+      "grad_norm": 6.411891937255859,
+      "learning_rate": 0.0001964037497738238,
+      "loss": 2.9513558959960937,
+      "step": 2350
+    },
+    {
+      "epoch": 1.14347616754935,
+      "grad_norm": 6.806637763977051,
+      "learning_rate": 0.00019629940762555336,
+      "loss": 2.8823919677734375,
+      "step": 2375
+    },
+    {
+      "epoch": 1.1555127587867116,
+      "grad_norm": 5.565702438354492,
+      "learning_rate": 0.0001961936018679537,
+      "loss": 2.8928656005859374,
+      "step": 2400
+    },
+    {
+      "epoch": 1.1675493500240732,
+      "grad_norm": 5.8602294921875,
+      "learning_rate": 0.00019608633410911667,
+      "loss": 2.9296820068359377,
+      "step": 2425
+    },
+    {
+      "epoch": 1.1795859412614347,
+      "grad_norm": 7.057725429534912,
+      "learning_rate": 0.0001959776059793544,
+      "loss": 2.926005859375,
+      "step": 2450
+    },
+    {
+      "epoch": 1.1916225324987963,
+      "grad_norm": 6.5474629402160645,
+      "learning_rate": 0.00019586741913117458,
+      "loss": 2.8745428466796876,
+      "step": 2475
+    },
+    {
+      "epoch": 1.2036591237361578,
+      "grad_norm": 5.752487659454346,
+      "learning_rate": 0.0001957557752392551,
+      "loss": 2.9059454345703126,
+      "step": 2500
+    },
+    {
+      "epoch": 1.2156957149735195,
+      "grad_norm": 6.054665565490723,
+      "learning_rate": 0.0001956426760004189,
+      "loss": 2.9109234619140625,
+      "step": 2525
+    },
+    {
+      "epoch": 1.2277323062108811,
+      "grad_norm": 5.9463276863098145,
+      "learning_rate": 0.0001955281231336081,
+      "loss": 2.8662490844726562,
+      "step": 2550
+    },
+    {
+      "epoch": 1.2397688974482426,
+      "grad_norm": 7.36589241027832,
+      "learning_rate": 0.00019541211837985768,
+      "loss": 2.9000286865234375,
+      "step": 2575
+    },
+    {
+      "epoch": 1.2518054886856043,
+      "grad_norm": 6.027646064758301,
+      "learning_rate": 0.0001952946635022693,
+      "loss": 2.8943441772460936,
+      "step": 2600
+    },
+    {
+      "epoch": 1.263842079922966,
+      "grad_norm": 5.916275978088379,
+      "learning_rate": 0.00019517576028598425,
+      "loss": 2.899263916015625,
+      "step": 2625
+    },
+    {
+      "epoch": 1.2758786711603274,
+      "grad_norm": 6.0202789306640625,
+      "learning_rate": 0.00019505541053815653,
+      "loss": 2.925803527832031,
+      "step": 2650
+    },
+    {
+      "epoch": 1.287915262397689,
+      "grad_norm": 6.8593339920043945,
+      "learning_rate": 0.00019493361608792528,
+      "loss": 2.9056402587890626,
+      "step": 2675
+    },
+    {
+      "epoch": 1.2999518536350505,
+      "grad_norm": 5.5273661613464355,
+      "learning_rate": 0.00019481037878638695,
+      "loss": 2.877239990234375,
+      "step": 2700
+    },
+    {
+      "epoch": 1.3119884448724122,
+      "grad_norm": 7.119853973388672,
+      "learning_rate": 0.0001946857005065672,
+      "loss": 2.85099853515625,
+      "step": 2725
+    },
+    {
+      "epoch": 1.3240250361097736,
+      "grad_norm": 5.379928112030029,
+      "learning_rate": 0.0001945595831433925,
+      "loss": 2.9064642333984376,
+      "step": 2750
+    },
+    {
+      "epoch": 1.3360616273471353,
+      "grad_norm": 6.6639604568481445,
+      "learning_rate": 0.00019443202861366124,
+      "loss": 2.882684326171875,
+      "step": 2775
+    },
+    {
+      "epoch": 1.348098218584497,
+      "grad_norm": 6.571842193603516,
+      "learning_rate": 0.00019430303885601465,
+      "loss": 2.933927307128906,
+      "step": 2800
+    },
+    {
+      "epoch": 1.3601348098218584,
+      "grad_norm": 6.048442363739014,
+      "learning_rate": 0.00019417261583090723,
+      "loss": 2.8845724487304687,
+      "step": 2825
+    },
+    {
+      "epoch": 1.37217140105922,
+      "grad_norm": 6.913212776184082,
+      "learning_rate": 0.00019404076152057713,
+      "loss": 2.92847412109375,
+      "step": 2850
+    },
+    {
+      "epoch": 1.3842079922965815,
+      "grad_norm": 6.1456122398376465,
+      "learning_rate": 0.0001939074779290159,
+      "loss": 2.864238586425781,
+      "step": 2875
+    },
+    {
+      "epoch": 1.3962445835339432,
+      "grad_norm": 6.314334869384766,
+      "learning_rate": 0.00019377276708193802,
+      "loss": 2.93041015625,
+      "step": 2900
+    },
+    {
+      "epoch": 1.4082811747713047,
+      "grad_norm": 6.124015808105469,
+      "learning_rate": 0.00019363663102675024,
+      "loss": 2.8582913208007814,
+      "step": 2925
+    },
+    {
+      "epoch": 1.4203177660086663,
+      "grad_norm": 5.999044895172119,
+      "learning_rate": 0.0001934990718325203,
+      "loss": 2.9177670288085937,
+      "step": 2950
+    },
+    {
+      "epoch": 1.432354357246028,
+      "grad_norm": 5.440136432647705,
+      "learning_rate": 0.00019336009158994562,
+      "loss": 2.942555236816406,
+      "step": 2975
+    },
+    {
+      "epoch": 1.4443909484833894,
+      "grad_norm": 6.6183929443359375,
+      "learning_rate": 0.00019321969241132138,
+      "loss": 2.93353515625,
+      "step": 3000
+    },
+    {
+      "epoch": 1.4564275397207511,
+      "grad_norm": 6.443948268890381,
+      "learning_rate": 0.00019307787643050857,
+      "loss": 2.914281311035156,
+      "step": 3025
+    },
+    {
+      "epoch": 1.4684641309581128,
+      "grad_norm": 6.57510232925415,
+      "learning_rate": 0.0001929346458029015,
+      "loss": 2.919800109863281,
+      "step": 3050
+    },
+    {
+      "epoch": 1.4805007221954742,
+      "grad_norm": 5.723294734954834,
+      "learning_rate": 0.00019279000270539497,
+      "loss": 2.875799255371094,
+      "step": 3075
+    },
+    {
+      "epoch": 1.4925373134328357,
+      "grad_norm": 6.246444225311279,
+      "learning_rate": 0.0001926439493363513,
+      "loss": 2.853829040527344,
+      "step": 3100
+    },
+    {
+      "epoch": 1.5045739046701974,
+      "grad_norm": 5.6569743156433105,
+      "learning_rate": 0.00019249648791556684,
+      "loss": 2.845904541015625,
+      "step": 3125
+    },
+    {
+      "epoch": 1.516610495907559,
+      "grad_norm": 5.917844295501709,
+      "learning_rate": 0.00019234762068423824,
+      "loss": 2.8845635986328126,
+      "step": 3150
+    },
+    {
+      "epoch": 1.5286470871449205,
+      "grad_norm": 7.2379302978515625,
+      "learning_rate": 0.00019219734990492843,
+      "loss": 2.8919659423828126,
+      "step": 3175
+    },
+    {
+      "epoch": 1.5406836783822822,
+      "grad_norm": 6.368435859680176,
+      "learning_rate": 0.00019204567786153213,
+      "loss": 2.8638519287109374,
+      "step": 3200
+    },
+    {
+      "epoch": 1.5527202696196438,
+      "grad_norm": 6.432679176330566,
+      "learning_rate": 0.0001918926068592414,
+      "loss": 2.8769232177734376,
+      "step": 3225
+    },
+    {
+      "epoch": 1.5647568608570053,
+      "grad_norm": 6.268744945526123,
+      "learning_rate": 0.00019173813922451015,
+      "loss": 2.88654296875,
+      "step": 3250
+    },
+    {
+      "epoch": 1.5767934520943667,
+      "grad_norm": 5.5019683837890625,
+      "learning_rate": 0.00019158227730501932,
+      "loss": 2.8765948486328123,
+      "step": 3275
+    },
+    {
+      "epoch": 1.5888300433317286,
+      "grad_norm": 5.808835983276367,
+      "learning_rate": 0.00019142502346964068,
+      "loss": 2.8841793823242186,
+      "step": 3300
+    },
+    {
+      "epoch": 1.60086663456909,
+      "grad_norm": 5.601189613342285,
+      "learning_rate": 0.00019126638010840126,
+      "loss": 2.90692138671875,
+      "step": 3325
+    },
+    {
+      "epoch": 1.6129032258064515,
+      "grad_norm": 6.727347373962402,
+      "learning_rate": 0.00019110634963244677,
+      "loss": 2.8471392822265624,
+      "step": 3350
+    },
+    {
+      "epoch": 1.6249398170438132,
+      "grad_norm": 5.520786762237549,
+      "learning_rate": 0.000190944934474005,
+      "loss": 2.825626220703125,
+      "step": 3375
+    },
+    {
+      "epoch": 1.6369764082811749,
+      "grad_norm": 5.6855244636535645,
+      "learning_rate": 0.0001907821370863489,
+      "loss": 2.85347900390625,
+      "step": 3400
+    },
+    {
+      "epoch": 1.6490129995185363,
+      "grad_norm": 6.275842189788818,
+      "learning_rate": 0.0001906179599437593,
+      "loss": 2.8564202880859373,
+      "step": 3425
+    },
+    {
+      "epoch": 1.661049590755898,
+      "grad_norm": 6.627660751342773,
+      "learning_rate": 0.0001904524055414872,
+      "loss": 2.88417236328125,
+      "step": 3450
+    },
+    {
+      "epoch": 1.6730861819932596,
+      "grad_norm": 5.4954609870910645,
+      "learning_rate": 0.00019028547639571598,
+      "loss": 2.883042297363281,
+      "step": 3475
+    },
+    {
+      "epoch": 1.685122773230621,
+      "grad_norm": 6.2831034660339355,
+      "learning_rate": 0.00019011717504352312,
+      "loss": 2.89915771484375,
+      "step": 3500
+    },
+    {
+      "epoch": 1.6971593644679825,
+      "grad_norm": 6.845497131347656,
+      "learning_rate": 0.00018994750404284155,
+      "loss": 2.81878173828125,
+      "step": 3525
+    },
+    {
+      "epoch": 1.7091959557053442,
+      "grad_norm": 7.301944732666016,
+      "learning_rate": 0.00018977646597242087,
+      "loss": 2.8858172607421877,
+      "step": 3550
+    },
+    {
+      "epoch": 1.7212325469427059,
+      "grad_norm": 7.100738525390625,
+      "learning_rate": 0.00018960406343178815,
+      "loss": 2.8665240478515623,
+      "step": 3575
+    },
+    {
+      "epoch": 1.7332691381800673,
+      "grad_norm": 6.5114593505859375,
+      "learning_rate": 0.00018943029904120836,
+      "loss": 2.8382354736328126,
+      "step": 3600
+    },
+    {
+      "epoch": 1.745305729417429,
+      "grad_norm": 5.751381874084473,
+      "learning_rate": 0.00018925517544164468,
+      "loss": 2.868603515625,
+      "step": 3625
+    },
+    {
+      "epoch": 1.7573423206547907,
+      "grad_norm": 5.941408634185791,
+      "learning_rate": 0.00018907869529471814,
+      "loss": 2.8781854248046876,
+      "step": 3650
+    },
+    {
+      "epoch": 1.7693789118921521,
+      "grad_norm": 5.040097236633301,
+      "learning_rate": 0.00018890086128266735,
+      "loss": 2.90862060546875,
+      "step": 3675
+    },
+    {
+      "epoch": 1.7814155031295136,
+      "grad_norm": 7.4268341064453125,
+      "learning_rate": 0.0001887216761083077,
+      "loss": 2.870086364746094,
+      "step": 3700
+    },
+    {
+      "epoch": 1.7934520943668752,
+      "grad_norm": 5.456490993499756,
+      "learning_rate": 0.00018854114249499025,
+      "loss": 2.8436822509765625,
+      "step": 3725
+    },
+    {
+      "epoch": 1.805488685604237,
+      "grad_norm": 6.014543533325195,
+      "learning_rate": 0.00018835926318656033,
+      "loss": 2.8642138671875,
+      "step": 3750
+    },
+    {
+      "epoch": 1.8175252768415984,
+      "grad_norm": 6.103621482849121,
+      "learning_rate": 0.00018817604094731582,
+      "loss": 2.8625579833984376,
+      "step": 3775
+    },
+    {
+      "epoch": 1.82956186807896,
+      "grad_norm": 6.694746494293213,
+      "learning_rate": 0.0001879914785619653,
+      "loss": 2.8301632690429686,
+      "step": 3800
+    },
+    {
+      "epoch": 1.8415984593163217,
+      "grad_norm": 5.783210754394531,
+      "learning_rate": 0.00018780557883558543,
+      "loss": 2.8412396240234377,
+      "step": 3825
+    },
+    {
+      "epoch": 1.8536350505536832,
+      "grad_norm": 6.31929349899292,
+      "learning_rate": 0.00018761834459357863,
+      "loss": 2.8531256103515625,
+      "step": 3850
+    },
+    {
+      "epoch": 1.8656716417910446,
+      "grad_norm": 5.855931282043457,
+      "learning_rate": 0.00018742977868162992,
+      "loss": 2.8694882202148437,
+      "step": 3875
+    },
+    {
+      "epoch": 1.8777082330284065,
+      "grad_norm": 5.388266563415527,
+      "learning_rate": 0.00018723988396566376,
+      "loss": 2.848956298828125,
+      "step": 3900
+    },
+    {
+      "epoch": 1.889744824265768,
+      "grad_norm": 7.263085842132568,
+      "learning_rate": 0.0001870486633318005,
+      "loss": 2.8653179931640627,
+      "step": 3925
+    },
+    {
+      "epoch": 1.9017814155031294,
+      "grad_norm": 5.926449298858643,
+      "learning_rate": 0.00018685611968631243,
+      "loss": 2.8881048583984374,
+      "step": 3950
+    },
+    {
+      "epoch": 1.913818006740491,
+      "grad_norm": 6.0646820068359375,
+      "learning_rate": 0.0001866622559555797,
+      "loss": 2.8977716064453123,
+      "step": 3975
+    },
+    {
+      "epoch": 1.9258545979778527,
+      "grad_norm": 5.905919075012207,
+      "learning_rate": 0.0001864670750860459,
+      "loss": 2.901802673339844,
+      "step": 4000
+    },
+    {
+      "epoch": 1.9378911892152142,
+      "grad_norm": 5.82496976852417,
+      "learning_rate": 0.00018627058004417303,
+      "loss": 2.8932440185546877,
+      "step": 4025
+    },
+    {
+      "epoch": 1.9499277804525759,
+      "grad_norm": 7.004024028778076,
+      "learning_rate": 0.00018607277381639672,
+      "loss": 2.8420306396484376,
+      "step": 4050
+    },
+    {
+      "epoch": 1.9619643716899375,
+      "grad_norm": 7.782563209533691,
+      "learning_rate": 0.0001858736594090806,
+      "loss": 2.901356201171875,
+      "step": 4075
+    },
+    {
+      "epoch": 1.974000962927299,
+      "grad_norm": 6.040978908538818,
+      "learning_rate": 0.00018567323984847075,
+      "loss": 2.8719561767578123,
+      "step": 4100
+    },
+    {
+      "epoch": 1.9860375541646604,
+      "grad_norm": 5.953182220458984,
+      "learning_rate": 0.00018547151818064966,
+      "loss": 2.86795166015625,
+      "step": 4125
+    },
+    {
+      "epoch": 1.998074145402022,
+      "grad_norm": 5.744591236114502,
+      "learning_rate": 0.00018526849747148996,
+      "loss": 2.860693359375,
+      "step": 4150
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 2.8608288764953613,
+      "eval_runtime": 4.4941,
+      "eval_samples_per_second": 333.771,
+      "eval_steps_per_second": 10.458,
+      "step": 4154
+    },
+    {
+      "epoch": 2.0101107366393838,
+      "grad_norm": 6.642997741699219,
+      "learning_rate": 0.0001850641808066077,
+      "loss": 2.6906396484375,
+      "step": 4175
+    },
+    {
+      "epoch": 2.0221473278767452,
+      "grad_norm": 6.480685234069824,
+      "learning_rate": 0.00018485857129131564,
+      "loss": 2.722727966308594,
+      "step": 4200
+    },
+    {
+      "epoch": 2.0341839191141067,
+      "grad_norm": 6.283753395080566,
+      "learning_rate": 0.00018465167205057597,
+      "loss": 2.683138732910156,
+      "step": 4225
+    },
+    {
+      "epoch": 2.0462205103514686,
+      "grad_norm": 5.976901531219482,
+      "learning_rate": 0.00018444348622895275,
+      "loss": 2.7119317626953126,
+      "step": 4250
+    },
+    {
+      "epoch": 2.05825710158883,
+      "grad_norm": 5.369650840759277,
+      "learning_rate": 0.00018423401699056423,
+      "loss": 2.71682861328125,
+      "step": 4275
+    },
+    {
+      "epoch": 2.0702936928261915,
+      "grad_norm": 5.941702842712402,
+      "learning_rate": 0.0001840232675190346,
+      "loss": 2.7161077880859374,
+      "step": 4300
+    },
+    {
+      "epoch": 2.0823302840635534,
+      "grad_norm": 6.431283950805664,
+      "learning_rate": 0.0001838112410174459,
+      "loss": 2.7145794677734374,
+      "step": 4325
+    },
+    {
+      "epoch": 2.094366875300915,
+      "grad_norm": 5.800447940826416,
+      "learning_rate": 0.000183597940708289,
+      "loss": 2.675470886230469,
+      "step": 4350
+    },
+    {
+      "epoch": 2.1064034665382763,
+      "grad_norm": 5.919703960418701,
+      "learning_rate": 0.00018338336983341475,
+      "loss": 2.70038818359375,
+      "step": 4375
+    },
+    {
+      "epoch": 2.118440057775638,
+      "grad_norm": 6.257649898529053,
+      "learning_rate": 0.0001831675316539849,
+      "loss": 2.69438720703125,
+      "step": 4400
+    },
+    {
+      "epoch": 2.1304766490129996,
+      "grad_norm": 5.656405925750732,
+      "learning_rate": 0.00018295042945042217,
+      "loss": 2.7394085693359376,
+      "step": 4425
+    },
+    {
+      "epoch": 2.142513240250361,
+      "grad_norm": 6.51097297668457,
+      "learning_rate": 0.0001827320665223607,
+      "loss": 2.763361511230469,
+      "step": 4450
+    },
+    {
+      "epoch": 2.1545498314877225,
+      "grad_norm": 6.375082969665527,
+      "learning_rate": 0.00018251244618859583,
+      "loss": 2.730120849609375,
+      "step": 4475
+    },
+    {
+      "epoch": 2.1665864227250844,
+      "grad_norm": 6.981193542480469,
+      "learning_rate": 0.00018229157178703342,
+      "loss": 2.71025146484375,
+      "step": 4500
+    },
+    {
+      "epoch": 2.178623013962446,
+      "grad_norm": 6.799978733062744,
+      "learning_rate": 0.00018206944667463953,
+      "loss": 2.68072265625,
+      "step": 4525
+    },
+    {
+      "epoch": 2.1906596051998073,
+      "grad_norm": 6.913100242614746,
+      "learning_rate": 0.0001818460742273891,
+      "loss": 2.73613037109375,
+      "step": 4550
+    },
+    {
+      "epoch": 2.202696196437169,
+      "grad_norm": 6.257125377655029,
+      "learning_rate": 0.00018162145784021466,
+      "loss": 2.756729736328125,
+      "step": 4575
+    },
+    {
+      "epoch": 2.2147327876745306,
+      "grad_norm": 5.430022239685059,
+      "learning_rate": 0.00018139560092695495,
+      "loss": 2.709376220703125,
+      "step": 4600
+    },
+    {
+      "epoch": 2.226769378911892,
+      "grad_norm": 6.211202621459961,
+      "learning_rate": 0.00018116850692030272,
+      "loss": 2.7708380126953127,
+      "step": 4625
+    },
+    {
+      "epoch": 2.2388059701492535,
+      "grad_norm": 6.597946643829346,
+      "learning_rate": 0.0001809401792717528,
+      "loss": 2.701777038574219,
+      "step": 4650
+    },
+    {
+      "epoch": 2.2508425613866154,
+      "grad_norm": 5.942161560058594,
+      "learning_rate": 0.0001807106214515496,
+      "loss": 2.6990762329101563,
+      "step": 4675
+    },
+    {
+      "epoch": 2.262879152623977,
+      "grad_norm": 7.107296466827393,
+      "learning_rate": 0.00018047983694863425,
+      "loss": 2.741151123046875,
+      "step": 4700
+    },
+    {
+      "epoch": 2.2749157438613383,
+      "grad_norm": 6.462876319885254,
+      "learning_rate": 0.00018024782927059167,
+      "loss": 2.7986865234375,
+      "step": 4725
+    },
+    {
+      "epoch": 2.2869523350987,
+      "grad_norm": 5.901287078857422,
+      "learning_rate": 0.0001800146019435973,
+      "loss": 2.723853759765625,
+      "step": 4750
+    },
+    {
+      "epoch": 2.2989889263360617,
+      "grad_norm": 6.557262420654297,
+      "learning_rate": 0.00017978015851236338,
+      "loss": 2.731398620605469,
+      "step": 4775
+    },
+    {
+      "epoch": 2.311025517573423,
+      "grad_norm": 6.787270545959473,
+      "learning_rate": 0.00017954450254008516,
+      "loss": 2.7483465576171877,
+      "step": 4800
+    },
+    {
+      "epoch": 2.3230621088107846,
+      "grad_norm": 7.6440887451171875,
+      "learning_rate": 0.0001793076376083868,
+      "loss": 2.7426837158203123,
+      "step": 4825
+    },
+    {
+      "epoch": 2.3350987000481465,
+      "grad_norm": 5.517909049987793,
+      "learning_rate": 0.00017906956731726675,
+      "loss": 2.718466796875,
+      "step": 4850
+    },
+    {
+      "epoch": 2.347135291285508,
+      "grad_norm": 5.788064956665039,
+      "learning_rate": 0.00017883029528504324,
+      "loss": 2.699304504394531,
+      "step": 4875
+    },
+    {
+      "epoch": 2.3591718825228694,
+      "grad_norm": 6.685894012451172,
+      "learning_rate": 0.00017858982514829913,
+      "loss": 2.743288269042969,
+      "step": 4900
+    },
+    {
+      "epoch": 2.3712084737602313,
+      "grad_norm": 5.696547031402588,
+      "learning_rate": 0.00017834816056182677,
+      "loss": 2.7296527099609373,
+      "step": 4925
+    },
+    {
+      "epoch": 2.3832450649975927,
+      "grad_norm": 6.552695274353027,
+      "learning_rate": 0.00017810530519857235,
+      "loss": 2.7500262451171875,
+      "step": 4950
+    },
+    {
+      "epoch": 2.395281656234954,
+      "grad_norm": 5.272420883178711,
+      "learning_rate": 0.00017786126274958015,
+      "loss": 2.7251507568359377,
+      "step": 4975
+    },
+    {
+      "epoch": 2.4073182474723156,
+      "grad_norm": 5.337221145629883,
+      "learning_rate": 0.00017761603692393636,
+      "loss": 2.794324951171875,
+      "step": 5000
+    },
+    {
+      "epoch": 2.4193548387096775,
+      "grad_norm": 6.288032531738281,
+      "learning_rate": 0.00017736963144871276,
+      "loss": 2.73385498046875,
+      "step": 5025
+    },
+    {
+      "epoch": 2.431391429947039,
+      "grad_norm": 6.103067398071289,
+      "learning_rate": 0.00017712205006891013,
+      "loss": 2.7030935668945313,
+      "step": 5050
+    },
+    {
+      "epoch": 2.4434280211844004,
+      "grad_norm": 5.142453670501709,
+      "learning_rate": 0.0001768732965474012,
+      "loss": 2.7068914794921874,
+      "step": 5075
+    },
+    {
+      "epoch": 2.4554646124217623,
+      "grad_norm": 6.099581241607666,
+      "learning_rate": 0.00017662337466487355,
+      "loss": 2.7641656494140623,
+      "step": 5100
+    },
+    {
+      "epoch": 2.4675012036591237,
+      "grad_norm": 6.829136848449707,
+      "learning_rate": 0.00017637228821977213,
+      "loss": 2.7043777465820313,
+      "step": 5125
+    },
+    {
+      "epoch": 2.479537794896485,
+      "grad_norm": 5.465605735778809,
+      "learning_rate": 0.00017612004102824153,
+      "loss": 2.7184182739257814,
+      "step": 5150
+    },
+    {
+      "epoch": 2.491574386133847,
+      "grad_norm": 6.038458347320557,
+      "learning_rate": 0.00017586663692406798,
+      "loss": 2.7635272216796873,
+      "step": 5175
+    },
+    {
+      "epoch": 2.5036109773712085,
+      "grad_norm": 6.614681720733643,
+      "learning_rate": 0.0001756120797586211,
+      "loss": 2.769054260253906,
+      "step": 5200
+    },
+    {
+      "epoch": 2.51564756860857,
+      "grad_norm": 5.841181755065918,
+      "learning_rate": 0.0001753563734007953,
+      "loss": 2.7224261474609377,
+      "step": 5225
+    },
+    {
+      "epoch": 2.527684159845932,
+      "grad_norm": 6.294864177703857,
+      "learning_rate": 0.00017509952173695107,
+      "loss": 2.7511309814453124,
+      "step": 5250
+    },
+    {
+      "epoch": 2.5397207510832933,
+      "grad_norm": 5.583828449249268,
+      "learning_rate": 0.00017484152867085583,
+      "loss": 2.7023577880859375,
+      "step": 5275
+    },
+    {
+      "epoch": 2.5517573423206548,
+      "grad_norm": 5.762092590332031,
+      "learning_rate": 0.00017458239812362462,
+      "loss": 2.7184487915039064,
+      "step": 5300
+    },
+    {
+      "epoch": 2.563793933558016,
+      "grad_norm": 7.2237677574157715,
+      "learning_rate": 0.0001743221340336606,
+      "loss": 2.746933898925781,
+      "step": 5325
+    },
+    {
+      "epoch": 2.575830524795378,
+      "grad_norm": 6.3399786949157715,
+      "learning_rate": 0.00017406074035659506,
+      "loss": 2.769549560546875,
+      "step": 5350
+    },
+    {
+      "epoch": 2.5878671160327396,
+      "grad_norm": 5.478464603424072,
+      "learning_rate": 0.0001737982210652273,
+      "loss": 2.738487548828125,
+      "step": 5375
+    },
+    {
+      "epoch": 2.599903707270101,
+      "grad_norm": 6.867932319641113,
+      "learning_rate": 0.00017353458014946442,
+      "loss": 2.7589501953125,
+      "step": 5400
+    },
+    {
+      "epoch": 2.611940298507463,
+      "grad_norm": 6.149633407592773,
+      "learning_rate": 0.00017326982161626044,
+      "loss": 2.7048358154296874,
+      "step": 5425
+    },
+    {
+      "epoch": 2.6239768897448243,
+      "grad_norm": 5.812656402587891,
+      "learning_rate": 0.00017300394948955558,
+      "loss": 2.7163870239257815,
+      "step": 5450
+    },
+    {
+      "epoch": 2.636013480982186,
+      "grad_norm": 8.48040771484375,
+      "learning_rate": 0.0001727369678102151,
+      "loss": 2.7666748046875,
+      "step": 5475
+    },
+    {
+      "epoch": 2.6480500722195472,
+      "grad_norm": 6.530597686767578,
+      "learning_rate": 0.00017246888063596774,
+      "loss": 2.683943176269531,
+      "step": 5500
+    },
+    {
+      "epoch": 2.660086663456909,
+      "grad_norm": 5.324703216552734,
+      "learning_rate": 0.00017219969204134412,
+      "loss": 2.756099853515625,
+      "step": 5525
+    },
+    {
+      "epoch": 2.6721232546942706,
+      "grad_norm": 5.394314289093018,
+      "learning_rate": 0.0001719294061176149,
+      "loss": 2.7569219970703127,
+      "step": 5550
+    },
+    {
+      "epoch": 2.684159845931632,
+      "grad_norm": 6.482044219970703,
+      "learning_rate": 0.00017165802697272854,
+      "loss": 2.7562155151367187,
+      "step": 5575
+    },
+    {
+      "epoch": 2.696196437168994,
+      "grad_norm": 5.999021530151367,
+      "learning_rate": 0.00017138555873124874,
+      "loss": 2.7514157104492187,
+      "step": 5600
+    },
+    {
+      "epoch": 2.7082330284063554,
+      "grad_norm": 6.616397380828857,
+      "learning_rate": 0.000171112005534292,
+      "loss": 2.709639892578125,
+      "step": 5625
+    },
+    {
+      "epoch": 2.720269619643717,
+      "grad_norm": 6.24782133102417,
+      "learning_rate": 0.0001708373715394645,
+      "loss": 2.7355551147460937,
+      "step": 5650
+    },
+    {
+      "epoch": 2.7323062108810783,
+      "grad_norm": 6.01416015625,
+      "learning_rate": 0.00017056166092079887,
+      "loss": 2.681227722167969,
+      "step": 5675
+    },
+    {
+      "epoch": 2.74434280211844,
+      "grad_norm": 5.480840682983398,
+      "learning_rate": 0.00017028487786869097,
+      "loss": 2.77478759765625,
+      "step": 5700
+    },
+    {
+      "epoch": 2.7563793933558016,
+      "grad_norm": 6.248979568481445,
+      "learning_rate": 0.00017000702658983603,
+      "loss": 2.71619140625,
+      "step": 5725
+    },
+    {
+      "epoch": 2.768415984593163,
+      "grad_norm": 6.709351062774658,
+      "learning_rate": 0.0001697281113071647,
+      "loss": 2.6932601928710938,
+      "step": 5750
+    },
+    {
+      "epoch": 2.780452575830525,
+      "grad_norm": 5.883296966552734,
+      "learning_rate": 0.00016944813625977895,
+      "loss": 2.7240725708007814,
+      "step": 5775
+    },
+    {
+      "epoch": 2.7924891670678864,
+      "grad_norm": 6.043994903564453,
+      "learning_rate": 0.00016916710570288767,
+      "loss": 2.730412902832031,
+      "step": 5800
+    },
+    {
+      "epoch": 2.804525758305248,
+      "grad_norm": 6.184457302093506,
+      "learning_rate": 0.0001688850239077419,
+      "loss": 2.7380224609375,
+      "step": 5825
+    },
+    {
+      "epoch": 2.8165623495426093,
+      "grad_norm": 6.543738842010498,
+      "learning_rate": 0.00016860189516156992,
+      "loss": 2.7130126953125,
+      "step": 5850
+    },
+    {
+      "epoch": 2.828598940779971,
+      "grad_norm": 6.47238826751709,
+      "learning_rate": 0.0001683177237675122,
+      "loss": 2.7353460693359377,
+      "step": 5875
+    },
+    {
+      "epoch": 2.8406355320173327,
+      "grad_norm": 5.968029975891113,
+      "learning_rate": 0.00016803251404455582,
+      "loss": 2.7125958251953124,
+      "step": 5900
+    },
+    {
+      "epoch": 2.8526721232546945,
+      "grad_norm": 6.4303717613220215,
+      "learning_rate": 0.00016774627032746904,
+      "loss": 2.710166015625,
+      "step": 5925
+    },
+    {
+      "epoch": 2.864708714492056,
+      "grad_norm": 5.7404093742370605,
+      "learning_rate": 0.0001674589969667353,
+      "loss": 2.7399676513671873,
+      "step": 5950
+    },
+    {
+      "epoch": 2.8767453057294174,
+      "grad_norm": 6.183196544647217,
+      "learning_rate": 0.00016717069832848698,
+      "loss": 2.7283358764648438,
+      "step": 5975
+    },
+    {
+      "epoch": 2.888781896966779,
+      "grad_norm": 5.792095184326172,
+      "learning_rate": 0.00016688137879443935,
+      "loss": 2.70381591796875,
+      "step": 6000
+    },
+    {
+      "epoch": 2.9008184882041403,
+      "grad_norm": 5.859752655029297,
+      "learning_rate": 0.00016659104276182365,
+      "loss": 2.7537738037109376,
+      "step": 6025
+    },
+    {
+      "epoch": 2.9128550794415022,
+      "grad_norm": 6.3950371742248535,
+      "learning_rate": 0.0001662996946433205,
+      "loss": 2.703232421875,
+      "step": 6050
+    },
+    {
+      "epoch": 2.9248916706788637,
+      "grad_norm": 6.948364734649658,
+      "learning_rate": 0.00016600733886699268,
+      "loss": 2.7639871215820313,
+      "step": 6075
+    },
+    {
+      "epoch": 2.9369282619162256,
+      "grad_norm": 6.1982197761535645,
+      "learning_rate": 0.0001657139798762179,
+      "loss": 2.733709411621094,
+      "step": 6100
+    },
+    {
+      "epoch": 2.948964853153587,
+      "grad_norm": 6.387999534606934,
+      "learning_rate": 0.00016541962212962127,
+      "loss": 2.7400689697265626,
+      "step": 6125
+    },
+    {
+      "epoch": 2.9610014443909485,
+      "grad_norm": 5.882162570953369,
+      "learning_rate": 0.00016512427010100748,
+      "loss": 2.7062359619140626,
+      "step": 6150
+    },
+    {
+      "epoch": 2.97303803562831,
+      "grad_norm": 7.070481777191162,
+      "learning_rate": 0.0001648279282792929,
+      "loss": 2.6777212524414065,
+      "step": 6175
+    },
+    {
+      "epoch": 2.9850746268656714,
+      "grad_norm": 5.267948150634766,
+      "learning_rate": 0.0001645306011684373,
+      "loss": 2.7145626831054686,
+      "step": 6200
+    },
+    {
+      "epoch": 2.9971112181030333,
+      "grad_norm": 6.219203948974609,
+      "learning_rate": 0.00016423229328737526,
+      "loss": 2.75112060546875,
+      "step": 6225
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 2.7634668350219727,
+      "eval_runtime": 4.1674,
+      "eval_samples_per_second": 359.936,
+      "eval_steps_per_second": 11.278,
+      "step": 6231
+    }
+  ],
+  "logging_steps": 25,
+  "max_steps": 20770,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}

checkpoint-6231/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1bb2c1c94aa0fdd3230268e016f0540025019dc2d2135c0ef925a101aa0459fa
+size 5265