Upload 11 files

adapter_config.json

@@ -21,11 +21,11 @@
  "revision": null,
  "target_modules": [
  "fc1",
- "dense",
  "q_proj",
- "v_proj",
  "fc2",
- "k_proj"
+ "dense",
+ "k_proj",
+ "v_proj"
  ],
  "task_type": "CAUSAL_LM",
  "use_dora": false,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:42b874a93d1bab2f97acafeeeb2678ccea95ae6104d49777039e9fb24070a081
+oid sha256:0ed1c2ec7bba2c3a47f4604e98ca91feaae4ae45eeeb381c0f377bff0b4dbe7c
 size 37774528

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:fe055b89dd8e17c1cd66b81060f6b2ca0dd1e1710b52e03801b2f0fb65f65a04
+oid sha256:910dc9e1ee19b05944cf3c1f6c5db091e12c734a1586ef6e956ddd847bdbe150
 size 75611578

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b8e4d3c512d3e7257319dac6f66f65df0ee5972b6230b8b0eabfac65e30eccea
+oid sha256:c8caf4be913ba0c7cfadaae796e6d56eb5e1583e0b176f7995b566b543235b55
 size 14244

tokenizer.json

@@ -1,6 +1,11 @@
 {
  "version": "1.0",
- "truncation": null,
+ "truncation": {
+ "direction": "Right",
+ "max_length": 2048,
+ "strategy": "LongestFirst",
+ "stride": 0
+ },
  "padding": null,
  "added_tokens": [
  {

trainer_state.json

@@ -8,3193 +8,645 @@
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
- {
- "epoch": 0.01098901098901099,
- "grad_norm": 3.238452911376953,
- "learning_rate": 0.00019956043956043958,
- "loss": 1.7274,
- "step": 100
- },
- {
- "epoch": 0.02197802197802198,
- "grad_norm": 2.8424575328826904,
- "learning_rate": 0.00019912087912087913,
- "loss": 1.7205,
- "step": 200
- },
- {
- "epoch": 0.03296703296703297,
- "grad_norm": 1.5604631900787354,
- "learning_rate": 0.0001986813186813187,
- "loss": 1.6773,
- "step": 300
- },
- {
- "epoch": 0.04395604395604396,
- "grad_norm": 2.338499069213867,
- "learning_rate": 0.00019824175824175827,
- "loss": 1.7067,
- "step": 400
- },
  {
  "epoch": 0.054945054945054944,
- "grad_norm": 2.3855156898498535,
+ "grad_norm": 1.780437707901001,
  "learning_rate": 0.0001978021978021978,
- "loss": 1.7492,
+ "loss": 1.5906,
  "step": 500
  },
- {
- "epoch": 0.06593406593406594,
- "grad_norm": 2.4433257579803467,
- "learning_rate": 0.00019736263736263738,
- "loss": 1.7072,
- "step": 600
- },
- {
- "epoch": 0.07692307692307693,
- "grad_norm": 2.889246702194214,
- "learning_rate": 0.00019692307692307696,
- "loss": 1.7447,
- "step": 700
- },
- {
- "epoch": 0.08791208791208792,
- "grad_norm": 3.244313955307007,
- "learning_rate": 0.00019648351648351647,
- "loss": 1.6999,
- "step": 800
- },
- {
- "epoch": 0.0989010989010989,
- "grad_norm": 2.247797727584839,
- "learning_rate": 0.00019604395604395604,
- "loss": 1.7091,
- "step": 900
- },
  {
  "epoch": 0.10989010989010989,
- "grad_norm": 3.4021923542022705,
+ "grad_norm": 2.4972989559173584,
  "learning_rate": 0.00019560439560439562,
- "loss": 1.7094,
+ "loss": 1.5616,
  "step": 1000
  },
- {
- "epoch": 0.12087912087912088,
- "grad_norm": 4.7427167892456055,
- "learning_rate": 0.00019516483516483516,
- "loss": 1.7151,
- "step": 1100
- },
- {
- "epoch": 0.13186813186813187,
- "grad_norm": 2.6960134506225586,
- "learning_rate": 0.00019472527472527473,
- "loss": 1.6853,
- "step": 1200
- },
- {
- "epoch": 0.14285714285714285,
- "grad_norm": 3.6055240631103516,
- "learning_rate": 0.0001942857142857143,
- "loss": 1.6763,
- "step": 1300
- },
- {
- "epoch": 0.15384615384615385,
- "grad_norm": 2.970062494277954,
- "learning_rate": 0.00019384615384615385,
- "loss": 1.7136,
- "step": 1400
- },
  {
  "epoch": 0.16483516483516483,
- "grad_norm": 3.8452701568603516,
+ "grad_norm": 1.6102287769317627,
  "learning_rate": 0.00019340659340659342,
- "loss": 1.7115,
+ "loss": 1.5549,
  "step": 1500
  },
- {
- "epoch": 0.17582417582417584,
- "grad_norm": 3.0747196674346924,
- "learning_rate": 0.000192967032967033,
- "loss": 1.7001,
- "step": 1600
- },
- {
- "epoch": 0.18681318681318682,
- "grad_norm": 2.819571018218994,
- "learning_rate": 0.00019252747252747254,
- "loss": 1.6939,
- "step": 1700
- },
- {
- "epoch": 0.1978021978021978,
- "grad_norm": 3.701860189437866,
- "learning_rate": 0.0001920879120879121,
- "loss": 1.6922,
- "step": 1800
- },
- {
- "epoch": 0.2087912087912088,
- "grad_norm": 2.7427585124969482,
- "learning_rate": 0.00019164835164835168,
- "loss": 1.717,
- "step": 1900
- },
  {
  "epoch": 0.21978021978021978,
- "grad_norm": 2.634605646133423,
+ "grad_norm": 1.6323072910308838,
  "learning_rate": 0.00019120879120879122,
- "loss": 1.6891,
+ "loss": 1.5446,
  "step": 2000
  },
- {
- "epoch": 0.23076923076923078,
- "grad_norm": 3.920029640197754,
- "learning_rate": 0.0001907692307692308,
- "loss": 1.7139,
- "step": 2100
- },
- {
- "epoch": 0.24175824175824176,
- "grad_norm": 2.9013421535491943,
- "learning_rate": 0.00019032967032967034,
- "loss": 1.6954,
- "step": 2200
- },
- {
- "epoch": 0.25274725274725274,
- "grad_norm": 2.9385688304901123,
- "learning_rate": 0.00018989010989010988,
- "loss": 1.7187,
- "step": 2300
- },
- {
- "epoch": 0.26373626373626374,
- "grad_norm": 3.105120897293091,
- "learning_rate": 0.00018945054945054945,
- "loss": 1.6687,
- "step": 2400
- },
  {
  "epoch": 0.27472527472527475,
- "grad_norm": 2.9849658012390137,
+ "grad_norm": 1.6228586435317993,
  "learning_rate": 0.00018901098901098903,
- "loss": 1.7133,
+ "loss": 1.5549,
  "step": 2500
  },
- {
- "epoch": 0.2857142857142857,
- "grad_norm": 2.239258289337158,
- "learning_rate": 0.00018857142857142857,
- "loss": 1.6609,
- "step": 2600
- },
- {
- "epoch": 0.2967032967032967,
- "grad_norm": 3.5539562702178955,
- "learning_rate": 0.00018813186813186814,
- "loss": 1.6778,
- "step": 2700
- },
- {
- "epoch": 0.3076923076923077,
- "grad_norm": 3.821715831756592,
- "learning_rate": 0.0001876923076923077,
- "loss": 1.6924,
- "step": 2800
- },
- {
- "epoch": 0.31868131868131866,
- "grad_norm": 2.939919948577881,
- "learning_rate": 0.00018725274725274726,
- "loss": 1.7103,
- "step": 2900
- },
  {
  "epoch": 0.32967032967032966,
- "grad_norm": 3.0582692623138428,
+ "grad_norm": 1.433160662651062,
  "learning_rate": 0.00018681318681318683,
- "loss": 1.7159,
+ "loss": 1.5521,
  "step": 3000
  },
- {
- "epoch": 0.34065934065934067,
- "grad_norm": 5.75909423828125,
- "learning_rate": 0.00018637362637362637,
- "loss": 1.6928,
- "step": 3100
- },
- {
- "epoch": 0.3516483516483517,
- "grad_norm": 4.105484485626221,
- "learning_rate": 0.00018593406593406595,
- "loss": 1.692,
- "step": 3200
- },
- {
- "epoch": 0.3626373626373626,
- "grad_norm": 2.9766411781311035,
- "learning_rate": 0.00018549450549450552,
- "loss": 1.6781,
- "step": 3300
- },
- {
- "epoch": 0.37362637362637363,
- "grad_norm": 3.47971510887146,
- "learning_rate": 0.00018505494505494506,
- "loss": 1.6668,
- "step": 3400
- },
  {
  "epoch": 0.38461538461538464,
- "grad_norm": 2.4754316806793213,
+ "grad_norm": 1.4724050760269165,
  "learning_rate": 0.00018461538461538463,
- "loss": 1.6662,
+ "loss": 1.5335,
  "step": 3500
  },
- {
- "epoch": 0.3956043956043956,
- "grad_norm": 2.745067834854126,
- "learning_rate": 0.00018417582417582418,
- "loss": 1.6739,
- "step": 3600
- },
- {
- "epoch": 0.4065934065934066,
- "grad_norm": 4.118983268737793,
- "learning_rate": 0.00018373626373626375,
- "loss": 1.6597,
- "step": 3700
- },
- {
- "epoch": 0.4175824175824176,
- "grad_norm": 3.8697848320007324,
- "learning_rate": 0.0001832967032967033,
- "loss": 1.6729,
- "step": 3800
- },
- {
- "epoch": 0.42857142857142855,
- "grad_norm": 7.602184772491455,
- "learning_rate": 0.00018285714285714286,
- "loss": 1.6892,
- "step": 3900
- },
  {
  "epoch": 0.43956043956043955,
- "grad_norm": 2.534014940261841,
+ "grad_norm": 1.2267966270446777,
  "learning_rate": 0.0001824175824175824,
- "loss": 1.6508,
+ "loss": 1.5359,
  "step": 4000
  },
- {
- "epoch": 0.45054945054945056,
- "grad_norm": 3.064992904663086,
- "learning_rate": 0.00018197802197802198,
- "loss": 1.6786,
- "step": 4100
- },
- {
- "epoch": 0.46153846153846156,
- "grad_norm": 3.288447141647339,
- "learning_rate": 0.00018153846153846155,
- "loss": 1.6594,
- "step": 4200
- },
- {
- "epoch": 0.4725274725274725,
- "grad_norm": 3.2647299766540527,
- "learning_rate": 0.0001810989010989011,
- "loss": 1.6593,
- "step": 4300
- },
- {
- "epoch": 0.4835164835164835,
- "grad_norm": 2.6522679328918457,
- "learning_rate": 0.00018065934065934067,
- "loss": 1.6682,
- "step": 4400
- },
  {
  "epoch": 0.4945054945054945,
- "grad_norm": 4.231736660003662,
+ "grad_norm": 1.3739681243896484,
  "learning_rate": 0.00018021978021978024,
- "loss": 1.6947,
+ "loss": 1.5362,
  "step": 4500
  },
- {
- "epoch": 0.5054945054945055,
- "grad_norm": 5.3127264976501465,
- "learning_rate": 0.00017978021978021978,
- "loss": 1.683,
- "step": 4600
- },
- {
- "epoch": 0.5164835164835165,
- "grad_norm": 4.50441837310791,
- "learning_rate": 0.00017934065934065936,
- "loss": 1.6701,
- "step": 4700
- },
- {
- "epoch": 0.5274725274725275,
- "grad_norm": 3.104093074798584,
- "learning_rate": 0.00017890109890109893,
- "loss": 1.6509,
- "step": 4800
- },
- {
- "epoch": 0.5384615384615384,
- "grad_norm": 2.652597427368164,
- "learning_rate": 0.00017846153846153847,
- "loss": 1.6523,
- "step": 4900
- },
  {
  "epoch": 0.5494505494505495,
- "grad_norm": 3.463927984237671,
+ "grad_norm": 1.3636854887008667,
  "learning_rate": 0.00017802197802197802,
- "loss": 1.662,
+ "loss": 1.5257,
  "step": 5000
  },
- {
- "epoch": 0.5604395604395604,
- "grad_norm": 2.976457357406616,
- "learning_rate": 0.0001775824175824176,
- "loss": 1.6843,
- "step": 5100
- },
- {
- "epoch": 0.5714285714285714,
- "grad_norm": 3.4616947174072266,
- "learning_rate": 0.00017714285714285713,
- "loss": 1.6989,
- "step": 5200
- },
- {
- "epoch": 0.5824175824175825,
- "grad_norm": 3.0407989025115967,
- "learning_rate": 0.0001767032967032967,
- "loss": 1.6601,
- "step": 5300
- },
- {
- "epoch": 0.5934065934065934,
- "grad_norm": 3.5591070652008057,
- "learning_rate": 0.00017626373626373627,
- "loss": 1.6641,
- "step": 5400
- },
  {
  "epoch": 0.6043956043956044,
- "grad_norm": 3.256300210952759,
+ "grad_norm": 1.9858155250549316,
  "learning_rate": 0.00017582417582417582,
- "loss": 1.6641,
+ "loss": 1.5378,
  "step": 5500
  },
- {
- "epoch": 0.6153846153846154,
- "grad_norm": 5.115685939788818,
- "learning_rate": 0.0001753846153846154,
- "loss": 1.6617,
- "step": 5600
- },
- {
- "epoch": 0.6263736263736264,
- "grad_norm": 4.381234645843506,
- "learning_rate": 0.00017494505494505496,
- "loss": 1.6844,
- "step": 5700
- },
- {
- "epoch": 0.6373626373626373,
- "grad_norm": 3.331096887588501,
- "learning_rate": 0.0001745054945054945,
- "loss": 1.6842,
- "step": 5800
- },
- {
- "epoch": 0.6483516483516484,
- "grad_norm": 4.683435916900635,
- "learning_rate": 0.00017406593406593408,
- "loss": 1.6583,
- "step": 5900
- },
  {
  "epoch": 0.6593406593406593,
- "grad_norm": 3.517239809036255,
+ "grad_norm": 1.6531416177749634,
  "learning_rate": 0.00017362637362637365,
- "loss": 1.6367,
+ "loss": 1.5239,
  "step": 6000
  },
- {
- "epoch": 0.6703296703296703,
- "grad_norm": 3.5134270191192627,
- "learning_rate": 0.0001731868131868132,
- "loss": 1.6623,
- "step": 6100
- },
- {
- "epoch": 0.6813186813186813,
- "grad_norm": 3.438363790512085,
- "learning_rate": 0.00017274725274725276,
- "loss": 1.6458,
- "step": 6200
- },
- {
- "epoch": 0.6923076923076923,
- "grad_norm": 3.1804699897766113,
- "learning_rate": 0.00017230769230769234,
- "loss": 1.7014,
- "step": 6300
- },
- {
- "epoch": 0.7032967032967034,
- "grad_norm": 2.5642647743225098,
- "learning_rate": 0.00017186813186813185,
- "loss": 1.6356,
- "step": 6400
- },
  {
  "epoch": 0.7142857142857143,
- "grad_norm": 2.30749249458313,
+ "grad_norm": 1.638098120689392,
  "learning_rate": 0.00017142857142857143,
- "loss": 1.695,
+ "loss": 1.5331,
  "step": 6500
  },
- {
- "epoch": 0.7252747252747253,
- "grad_norm": 2.9326508045196533,
- "learning_rate": 0.000170989010989011,
- "loss": 1.6406,
- "step": 6600
- },
- {
- "epoch": 0.7362637362637363,
- "grad_norm": 3.912691593170166,
- "learning_rate": 0.00017054945054945054,
- "loss": 1.6604,
- "step": 6700
- },
- {
- "epoch": 0.7472527472527473,
- "grad_norm": 4.329016208648682,
- "learning_rate": 0.0001701098901098901,
- "loss": 1.7025,
- "step": 6800
- },
- {
- "epoch": 0.7582417582417582,
- "grad_norm": 4.046297073364258,
- "learning_rate": 0.00016967032967032968,
- "loss": 1.6578,
- "step": 6900
- },
  {
  "epoch": 0.7692307692307693,
- "grad_norm": 3.2432618141174316,
+ "grad_norm": 1.8147164583206177,
  "learning_rate": 0.00016923076923076923,
- "loss": 1.6281,
+ "loss": 1.5291,
  "step": 7000
  },
- {
- "epoch": 0.7802197802197802,
- "grad_norm": 4.125152587890625,
- "learning_rate": 0.0001687912087912088,
- "loss": 1.6808,
- "step": 7100
- },
- {
- "epoch": 0.7912087912087912,
- "grad_norm": 3.0072073936462402,
- "learning_rate": 0.00016835164835164837,
- "loss": 1.6935,
- "step": 7200
- },
- {
- "epoch": 0.8021978021978022,
- "grad_norm": 3.7964279651641846,
- "learning_rate": 0.00016791208791208792,
- "loss": 1.6772,
- "step": 7300
- },
- {
- "epoch": 0.8131868131868132,
- "grad_norm": 4.463648796081543,
- "learning_rate": 0.0001674725274725275,
- "loss": 1.6704,
- "step": 7400
- },
  {
  "epoch": 0.8241758241758241,
- "grad_norm": 4.678513050079346,
+ "grad_norm": 1.4330575466156006,
  "learning_rate": 0.00016703296703296706,
- "loss": 1.6339,
+ "loss": 1.5328,
  "step": 7500
  },
- {
- "epoch": 0.8351648351648352,
- "grad_norm": 4.4003214836120605,
- "learning_rate": 0.0001665934065934066,
- "loss": 1.6491,
- "step": 7600
- },
- {
- "epoch": 0.8461538461538461,
- "grad_norm": 2.7502541542053223,
- "learning_rate": 0.00016615384615384617,
- "loss": 1.6882,
- "step": 7700
- },
- {
- "epoch": 0.8571428571428571,
- "grad_norm": 3.120934247970581,
- "learning_rate": 0.00016571428571428575,
- "loss": 1.6457,
- "step": 7800
- },
- {
- "epoch": 0.8681318681318682,
- "grad_norm": 4.110687732696533,
- "learning_rate": 0.00016527472527472526,
- "loss": 1.6817,
- "step": 7900
- },
  {
  "epoch": 0.8791208791208791,
- "grad_norm": 3.5830562114715576,
+ "grad_norm": 1.3865350484848022,
  "learning_rate": 0.00016483516483516484,
- "loss": 1.6395,
+ "loss": 1.5266,
  "step": 8000
  },
- {
- "epoch": 0.8901098901098901,
- "grad_norm": 3.775438070297241,
- "learning_rate": 0.0001643956043956044,
- "loss": 1.6811,
- "step": 8100
- },
- {
- "epoch": 0.9010989010989011,
- "grad_norm": 3.679263114929199,
- "learning_rate": 0.00016395604395604395,
- "loss": 1.6827,
- "step": 8200
- },
- {
- "epoch": 0.9120879120879121,
- "grad_norm": 3.706937313079834,
- "learning_rate": 0.00016351648351648352,
- "loss": 1.6554,
- "step": 8300
- },
- {
- "epoch": 0.9230769230769231,
- "grad_norm": 3.64091420173645,
- "learning_rate": 0.0001630769230769231,
- "loss": 1.6496,
- "step": 8400
- },
  {
  "epoch": 0.9340659340659341,
- "grad_norm": 2.954841136932373,
+ "grad_norm": 1.8921843767166138,
  "learning_rate": 0.00016263736263736264,
- "loss": 1.6644,
+ "loss": 1.5323,
  "step": 8500
  },
- {
- "epoch": 0.945054945054945,
- "grad_norm": 3.5718305110931396,
- "learning_rate": 0.0001621978021978022,
- "loss": 1.6597,
- "step": 8600
- },
- {
- "epoch": 0.9560439560439561,
- "grad_norm": 3.4999473094940186,
- "learning_rate": 0.00016175824175824178,
- "loss": 1.6563,
- "step": 8700
- },
- {
- "epoch": 0.967032967032967,
- "grad_norm": 4.831845760345459,
- "learning_rate": 0.00016131868131868133,
- "loss": 1.645,
- "step": 8800
- },
- {
- "epoch": 0.978021978021978,
- "grad_norm": 2.883336067199707,
- "learning_rate": 0.0001608791208791209,
- "loss": 1.6712,
- "step": 8900
- },
  {
  "epoch": 0.989010989010989,
- "grad_norm": 4.344512462615967,
+ "grad_norm": 2.120455026626587,
  "learning_rate": 0.00016043956043956044,
- "loss": 1.6542,
+ "loss": 1.5254,
  "step": 9000
  },
- {
- "epoch": 1.0,
- "grad_norm": 3.8828916549682617,
- "learning_rate": 0.00016,
- "loss": 1.6529,
- "step": 9100
- },
- {
- "epoch": 1.010989010989011,
- "grad_norm": 3.094627857208252,
- "learning_rate": 0.00015956043956043958,
- "loss": 1.5436,
- "step": 9200
- },
- {
- "epoch": 1.021978021978022,
- "grad_norm": 3.886928081512451,
- "learning_rate": 0.00015912087912087913,
- "loss": 1.5088,
- "step": 9300
- },
- {
- "epoch": 1.032967032967033,
- "grad_norm": 2.9219202995300293,
- "learning_rate": 0.00015868131868131867,
- "loss": 1.567,
- "step": 9400
- },
  {
  "epoch": 1.043956043956044,
- "grad_norm": 3.28596568107605,
+ "grad_norm": 1.6023536920547485,
  "learning_rate": 0.00015824175824175824,
- "loss": 1.5649,
+ "loss": 1.4525,
  "step": 9500
  },
- {
- "epoch": 1.054945054945055,
- "grad_norm": 3.9126474857330322,
- "learning_rate": 0.00015780219780219782,
- "loss": 1.5968,
- "step": 9600
- },
- {
- "epoch": 1.065934065934066,
- "grad_norm": 4.429248332977295,
- "learning_rate": 0.00015736263736263736,
- "loss": 1.5518,
- "step": 9700
- },
- {
- "epoch": 1.0769230769230769,
- "grad_norm": 3.1986796855926514,
- "learning_rate": 0.00015692307692307693,
- "loss": 1.566,
- "step": 9800
- },
- {
- "epoch": 1.0879120879120878,
- "grad_norm": 7.024231910705566,
- "learning_rate": 0.00015648351648351648,
- "loss": 1.5857,
- "step": 9900
- },
  {
  "epoch": 1.098901098901099,
- "grad_norm": 3.3280653953552246,
+ "grad_norm": 1.4707773923873901,
  "learning_rate": 0.00015604395604395605,
- "loss": 1.5898,
+ "loss": 1.4538,
  "step": 10000
  },
- {
- "epoch": 1.10989010989011,
- "grad_norm": 4.216164588928223,
- "learning_rate": 0.00015560439560439562,
- "loss": 1.6087,
- "step": 10100
- },
- {
- "epoch": 1.120879120879121,
- "grad_norm": 4.030162334442139,
- "learning_rate": 0.00015516483516483516,
- "loss": 1.5605,
- "step": 10200
- },
- {
- "epoch": 1.1318681318681318,
- "grad_norm": 4.694043159484863,
- "learning_rate": 0.00015472527472527474,
- "loss": 1.5485,
- "step": 10300
- },
- {
- "epoch": 1.1428571428571428,
- "grad_norm": 4.568836688995361,
- "learning_rate": 0.0001542857142857143,
- "loss": 1.5422,
- "step": 10400
- },
  {
  "epoch": 1.1538461538461537,
- "grad_norm": 5.25291109085083,
+ "grad_norm": 1.7984991073608398,
  "learning_rate": 0.00015384615384615385,
- "loss": 1.5478,
+ "loss": 1.4421,
  "step": 10500
  },
- {
- "epoch": 1.164835164835165,
- "grad_norm": 5.639504432678223,
- "learning_rate": 0.00015340659340659342,
- "loss": 1.5752,
- "step": 10600
- },
- {
- "epoch": 1.1758241758241759,
- "grad_norm": 4.555628299713135,
- "learning_rate": 0.00015296703296703297,
- "loss": 1.5716,
- "step": 10700
- },
- {
- "epoch": 1.1868131868131868,
- "grad_norm": 4.831645488739014,
- "learning_rate": 0.00015252747252747254,
- "loss": 1.5895,
- "step": 10800
- },
- {
- "epoch": 1.1978021978021978,
- "grad_norm": 5.440847873687744,
- "learning_rate": 0.00015208791208791208,
- "loss": 1.5499,
- "step": 10900
- },
  {
  "epoch": 1.2087912087912087,
- "grad_norm": 3.4942378997802734,
+ "grad_norm": 1.70187509059906,
  "learning_rate": 0.00015164835164835165,
- "loss": 1.5743,
+ "loss": 1.4454,
  "step": 11000
  },
- {
- "epoch": 1.2197802197802199,
- "grad_norm": 4.695743560791016,
- "learning_rate": 0.0001512087912087912,
- "loss": 1.607,
- "step": 11100
- },
- {
- "epoch": 1.2307692307692308,
- "grad_norm": 3.870124101638794,
- "learning_rate": 0.00015076923076923077,
- "loss": 1.5658,
- "step": 11200
- },
- {
- "epoch": 1.2417582417582418,
- "grad_norm": 3.653447389602661,
- "learning_rate": 0.00015032967032967034,
- "loss": 1.5651,
- "step": 11300
- },
- {
- "epoch": 1.2527472527472527,
- "grad_norm": 3.547433853149414,
- "learning_rate": 0.0001498901098901099,
- "loss": 1.5628,
- "step": 11400
- },
  {
  "epoch": 1.2637362637362637,
- "grad_norm": 5.93261194229126,
+ "grad_norm": 2.5541532039642334,
  "learning_rate": 0.00014945054945054946,
- "loss": 1.5611,
+ "loss": 1.4512,
  "step": 11500
  },
- {
- "epoch": 1.2747252747252746,
- "grad_norm": 3.197800636291504,
- "learning_rate": 0.00014901098901098903,
- "loss": 1.5782,
- "step": 11600
- },
- {
- "epoch": 1.2857142857142856,
- "grad_norm": 3.305222988128662,
- "learning_rate": 0.00014857142857142857,
- "loss": 1.5998,
- "step": 11700
- },
- {
- "epoch": 1.2967032967032968,
- "grad_norm": 3.467773914337158,
- "learning_rate": 0.00014813186813186815,
- "loss": 1.5951,
- "step": 11800
- },
- {
- "epoch": 1.3076923076923077,
- "grad_norm": 4.193416595458984,
- "learning_rate": 0.00014769230769230772,
- "loss": 1.592,
- "step": 11900
- },
  {
  "epoch": 1.3186813186813187,
- "grad_norm": 4.387876510620117,
+ "grad_norm": 1.5920054912567139,
  "learning_rate": 0.00014725274725274726,
- "loss": 1.5502,
+ "loss": 1.4557,
  "step": 12000
  },
- {
- "epoch": 1.3296703296703296,
- "grad_norm": 3.4815001487731934,
- "learning_rate": 0.0001468131868131868,
- "loss": 1.5546,
- "step": 12100
- },
- {
- "epoch": 1.3406593406593408,
- "grad_norm": 8.390545845031738,
- "learning_rate": 0.00014637362637362638,
- "loss": 1.5886,
- "step": 12200
- },
- {
- "epoch": 1.3516483516483517,
- "grad_norm": 4.278797626495361,
- "learning_rate": 0.00014593406593406592,
- "loss": 1.5537,
- "step": 12300
- },
- {
- "epoch": 1.3626373626373627,
- "grad_norm": 4.196005821228027,
- "learning_rate": 0.0001454945054945055,
- "loss": 1.5602,
- "step": 12400
- },
  {
  "epoch": 1.3736263736263736,
- "grad_norm": 2.9898526668548584,
+ "grad_norm": 2.0435898303985596,
  "learning_rate": 0.00014505494505494506,
- "loss": 1.5804,
+ "loss": 1.4496,
  "step": 12500
  },
- {
- "epoch": 1.3846153846153846,
- "grad_norm": 3.346822500228882,
- "learning_rate": 0.0001446153846153846,
- "loss": 1.5658,
- "step": 12600
- },
- {
- "epoch": 1.3956043956043955,
- "grad_norm": 4.593723773956299,
- "learning_rate": 0.00014417582417582418,
- "loss": 1.5846,
- "step": 12700
- },
- {
- "epoch": 1.4065934065934065,
- "grad_norm": 4.338139057159424,
- "learning_rate": 0.00014373626373626375,
- "loss": 1.5682,
- "step": 12800
- },
- {
- "epoch": 1.4175824175824177,
- "grad_norm": 4.005186557769775,
- "learning_rate": 0.0001432967032967033,
- "loss": 1.5411,
- "step": 12900
- },
  {
  "epoch": 1.4285714285714286,
- "grad_norm": 3.662499189376831,
+ "grad_norm": 2.3308217525482178,
  "learning_rate": 0.00014285714285714287,
- "loss": 1.5364,
+ "loss": 1.437,
  "step": 13000
  },
- {
- "epoch": 1.4395604395604396,
- "grad_norm": 3.864833354949951,
- "learning_rate": 0.00014241758241758244,
- "loss": 1.5794,
- "step": 13100
- },
- {
- "epoch": 1.4505494505494505,
- "grad_norm": 5.973376750946045,
- "learning_rate": 0.00014197802197802198,
- "loss": 1.5585,
- "step": 13200
- },
- {
- "epoch": 1.4615384615384617,
- "grad_norm": 4.784486770629883,
- "learning_rate": 0.00014153846153846156,
- "loss": 1.5396,
- "step": 13300
- },
- {
- "epoch": 1.4725274725274726,
- "grad_norm": 3.6531691551208496,
- "learning_rate": 0.00014109890109890113,
- "loss": 1.5427,
- "step": 13400
- },
  {
  "epoch": 1.4835164835164836,
- "grad_norm": 4.697863578796387,
+ "grad_norm": 2.7433712482452393,
  "learning_rate": 0.00014065934065934067,
- "loss": 1.569,
+ "loss": 1.437,
  "step": 13500
  },
- {
- "epoch": 1.4945054945054945,
- "grad_norm": 4.377908706665039,
- "learning_rate": 0.00014021978021978022,
- "loss": 1.5393,
- "step": 13600
- },
- {
- "epoch": 1.5054945054945055,
- "grad_norm": 10.914487838745117,
- "learning_rate": 0.0001397802197802198,
- "loss": 1.564,
- "step": 13700
- },
- {
- "epoch": 1.5164835164835164,
- "grad_norm": 5.559396743774414,
- "learning_rate": 0.00013934065934065933,
- "loss": 1.5925,
- "step": 13800
- },
- {
- "epoch": 1.5274725274725274,
- "grad_norm": 4.534477710723877,
- "learning_rate": 0.0001389010989010989,
- "loss": 1.5753,
- "step": 13900
- },
  {
  "epoch": 1.5384615384615383,
- "grad_norm": 5.669466495513916,
+ "grad_norm": 2.6499412059783936,
  "learning_rate": 0.00013846153846153847,
- "loss": 1.5519,
+ "loss": 1.4419,
  "step": 14000
  },
- {
- "epoch": 1.5494505494505495,
- "grad_norm": 5.009109020233154,
- "learning_rate": 0.00013802197802197802,
- "loss": 1.5559,
- "step": 14100
- },
- {
- "epoch": 1.5604395604395604,
- "grad_norm": 3.9765381813049316,
- "learning_rate": 0.0001375824175824176,
- "loss": 1.5458,
- "step": 14200
- },
- {
- "epoch": 1.5714285714285714,
- "grad_norm": 5.045168399810791,
- "learning_rate": 0.00013714285714285716,
- "loss": 1.5594,
- "step": 14300
- },
- {
- "epoch": 1.5824175824175826,
- "grad_norm": 5.065695285797119,
- "learning_rate": 0.0001367032967032967,
- "loss": 1.5523,
- "step": 14400
- },
  {
  "epoch": 1.5934065934065935,
- "grad_norm": 7.845288276672363,
+ "grad_norm": 2.6177027225494385,
  "learning_rate": 0.00013626373626373628,
- "loss": 1.5768,
+ "loss": 1.4408,
  "step": 14500
  },
- {
- "epoch": 1.6043956043956045,
- "grad_norm": 3.814141273498535,
- "learning_rate": 0.00013582417582417585,
- "loss": 1.5937,
- "step": 14600
- },
- {
- "epoch": 1.6153846153846154,
- "grad_norm": 3.292379379272461,
- "learning_rate": 0.0001353846153846154,
- "loss": 1.5506,
- "step": 14700
- },
- {
- "epoch": 1.6263736263736264,
- "grad_norm": 3.32590389251709,
- "learning_rate": 0.00013494505494505497,
- "loss": 1.5501,
- "step": 14800
- },
- {
- "epoch": 1.6373626373626373,
- "grad_norm": 4.293301582336426,
- "learning_rate": 0.0001345054945054945,
- "loss": 1.5684,
- "step": 14900
- },
  {
  "epoch": 1.6483516483516483,
- "grad_norm": 4.547417163848877,
+ "grad_norm": 1.9484443664550781,
  "learning_rate": 0.00013406593406593405,
- "loss": 1.6011,
+ "loss": 1.4617,
  "step": 15000
  },
  {
- "epoch": 1.6593406593406592,
- "grad_norm": 8.499897956848145,
- "learning_rate": 0.00013362637362637363,
- "loss": 1.5612,
- "step": 15100
+ "epoch": 1.7032967032967035,
+ "grad_norm": 2.649055242538452,
+ "learning_rate": 0.00013186813186813188,
+ "loss": 1.4482,
+ "step": 15500
  },
  {
- "epoch": 1.6703296703296702,
- "grad_norm": 4.393691539764404,
- "learning_rate": 0.0001331868131868132,
- "loss": 1.5588,
- "step": 15200
+ "epoch": 1.7582417582417582,
+ "grad_norm": 1.9473294019699097,
+ "learning_rate": 0.0001296703296703297,
+ "loss": 1.4358,
+ "step": 16000
  },
  {
- "epoch": 1.6813186813186813,
- "grad_norm": 5.121108055114746,
- "learning_rate": 0.00013274725274725274,
- "loss": 1.5606,
- "step": 15300
+ "epoch": 1.8131868131868132,
+ "grad_norm": 2.2678744792938232,
+ "learning_rate": 0.00012747252747252746,
+ "loss": 1.4333,
+ "step": 16500
  },
  {
- "epoch": 1.6923076923076923,
- "grad_norm": 3.4361484050750732,
- "learning_rate": 0.0001323076923076923,
- "loss": 1.5746,
- "step": 15400
- },
- {
- "epoch": 1.7032967032967035,
- "grad_norm": 3.9485526084899902,
- "learning_rate": 0.00013186813186813188,
- "loss": 1.5811,
- "step": 15500
- },
- {
- "epoch": 1.7142857142857144,
- "grad_norm": 5.078463554382324,
- "learning_rate": 0.00013142857142857143,
- "loss": 1.5576,
- "step": 15600
- },
- {
- "epoch": 1.7252747252747254,
- "grad_norm": 3.7762668132781982,
- "learning_rate": 0.000130989010989011,
- "loss": 1.5502,
- "step": 15700
- },
- {
- "epoch": 1.7362637362637363,
- "grad_norm": 4.067686080932617,
- "learning_rate": 0.00013054945054945057,
- "loss": 1.5486,
- "step": 15800
- },
- {
- "epoch": 1.7472527472527473,
- "grad_norm": 3.324613094329834,
- "learning_rate": 0.00013010989010989012,
- "loss": 1.5786,
- "step": 15900
- },
- {
- "epoch": 1.7582417582417582,
- "grad_norm": 3.8743014335632324,
- "learning_rate": 0.0001296703296703297,
- "loss": 1.5341,
- "step": 16000
- },
- {
- "epoch": 1.7692307692307692,
- "grad_norm": 3.096376419067383,
- "learning_rate": 0.00012923076923076923,
- "loss": 1.5428,
- "step": 16100
- },
- {
- "epoch": 1.7802197802197801,
- "grad_norm": 5.382870197296143,
- "learning_rate": 0.0001287912087912088,
- "loss": 1.5364,
- "step": 16200
- },
- {
- "epoch": 1.791208791208791,
- "grad_norm": 7.159231185913086,
- "learning_rate": 0.00012835164835164837,
- "loss": 1.5559,
- "step": 16300
- },
- {
- "epoch": 1.8021978021978022,
- "grad_norm": 3.7771100997924805,
- "learning_rate": 0.00012791208791208792,
- "loss": 1.5576,
- "step": 16400
- },
- {
- "epoch": 1.8131868131868132,
- "grad_norm": 3.727796792984009,
- "learning_rate": 0.00012747252747252746,
- "loss": 1.5741,
- "step": 16500
- },
- {
- "epoch": 1.8241758241758241,
- "grad_norm": 4.148353576660156,
- "learning_rate": 0.00012703296703296704,
- "loss": 1.5618,
- "step": 16600
- },
- {
- "epoch": 1.8351648351648353,
- "grad_norm": 4.795083999633789,
- "learning_rate": 0.0001265934065934066,
- "loss": 1.5203,
- "step": 16700
- },
- {
- "epoch": 1.8461538461538463,
- "grad_norm": 4.537530422210693,
- "learning_rate": 0.00012615384615384615,
- "loss": 1.5464,
- "step": 16800
- },
- {
- "epoch": 1.8571428571428572,
- "grad_norm": 6.016224384307861,
- "learning_rate": 0.00012571428571428572,
- "loss": 1.5484,
- "step": 16900
- },
- {
- "epoch": 1.8681318681318682,
- "grad_norm": 4.713044166564941,
- "learning_rate": 0.00012527472527472527,
- "loss": 1.5361,
- "step": 17000
- },
- {
- "epoch": 1.879120879120879,
- "grad_norm": 4.3841094970703125,
- "learning_rate": 0.00012483516483516484,
- "loss": 1.5591,
- "step": 17100
- },
- {
- "epoch": 1.89010989010989,
- "grad_norm": 5.991598129272461,
- "learning_rate": 0.0001243956043956044,
- "loss": 1.5474,
- "step": 17200
- },
- {
- "epoch": 1.901098901098901,
- "grad_norm": 4.520913600921631,
- "learning_rate": 0.00012395604395604395,
- "loss": 1.5269,
- "step": 17300
- },
- {
- "epoch": 1.912087912087912,
- "grad_norm": 4.3360066413879395,
- "learning_rate": 0.00012351648351648353,
- "loss": 1.5665,
- "step": 17400
+ "epoch": 1.8681318681318682,
+ "grad_norm": 1.629854440689087,
+ "learning_rate": 0.00012527472527472527,
+ "loss": 1.4305,
+ "step": 17000
  },
  {
  "epoch": 1.9230769230769231,
- "grad_norm": 4.132837295532227,
+ "grad_norm": 2.318514823913574,
  "learning_rate": 0.0001230769230769231,
- "loss": 1.553,
+ "loss": 1.4292,
  "step": 17500
  },
- {
- "epoch": 1.934065934065934,
- "grad_norm": 3.756524085998535,
- "learning_rate": 0.00012263736263736264,
- "loss": 1.5602,
- "step": 17600
- },
- {
- "epoch": 1.945054945054945,
- "grad_norm": 3.8551275730133057,
- "learning_rate": 0.0001221978021978022,
- "loss": 1.55,
- "step": 17700
- },
- {
- "epoch": 1.9560439560439562,
- "grad_norm": 5.236588001251221,
- "learning_rate": 0.00012175824175824176,
- "loss": 1.5409,
- "step": 17800
- },
- {
- "epoch": 1.9670329670329672,
- "grad_norm": 4.134451389312744,
- "learning_rate": 0.00012131868131868132,
- "loss": 1.5126,
- "step": 17900
- },
  {
  "epoch": 1.978021978021978,
- "grad_norm": 4.225131034851074,
+ "grad_norm": 2.1276426315307617,
  "learning_rate": 0.00012087912087912087,
- "loss": 1.5824,
+ "loss": 1.435,
  "step": 18000
  },
- {
- "epoch": 1.989010989010989,
- "grad_norm": 4.457215309143066,
- "learning_rate": 0.00012043956043956045,
- "loss": 1.554,
- "step": 18100
- },
- {
- "epoch": 2.0,
- "grad_norm": 3.8229620456695557,
- "learning_rate": 0.00012,
- "loss": 1.5436,
- "step": 18200
- },
- {
- "epoch": 2.010989010989011,
- "grad_norm": 3.4859657287597656,
- "learning_rate": 0.00011956043956043956,
- "loss": 1.4374,
- "step": 18300
- },
- {
- "epoch": 2.021978021978022,
- "grad_norm": 4.995362758636475,
- "learning_rate": 0.00011912087912087912,
- "loss": 1.4391,
- "step": 18400
- },
  {
  "epoch": 2.032967032967033,
- "grad_norm": 3.7239787578582764,
+ "grad_norm": 2.3333852291107178,
  "learning_rate": 0.00011868131868131869,
- "loss": 1.4505,
+ "loss": 1.3761,
  "step": 18500
  },
- {
- "epoch": 2.043956043956044,
- "grad_norm": 2.1358206272125244,
- "learning_rate": 0.00011824175824175825,
- "loss": 1.4027,
- "step": 18600
- },
- {
- "epoch": 2.0549450549450547,
- "grad_norm": 6.7210187911987305,
- "learning_rate": 0.0001178021978021978,
- "loss": 1.4537,
- "step": 18700
- },
- {
- "epoch": 2.065934065934066,
- "grad_norm": 3.9406368732452393,
- "learning_rate": 0.00011736263736263738,
- "loss": 1.433,
- "step": 18800
- },
- {
- "epoch": 2.076923076923077,
- "grad_norm": 4.559463977813721,
- "learning_rate": 0.00011692307692307694,
- "loss": 1.4321,
- "step": 18900
- },
  {
  "epoch": 2.087912087912088,
- "grad_norm": 4.0497236251831055,
+ "grad_norm": 3.305957078933716,
  "learning_rate": 0.0001164835164835165,
- "loss": 1.456,
+ "loss": 1.3318,
  "step": 19000
  },
- {
- "epoch": 2.098901098901099,
- "grad_norm": 4.5423760414123535,
- "learning_rate": 0.00011604395604395607,
- "loss": 1.4326,
- "step": 19100
- },
- {
- "epoch": 2.10989010989011,
- "grad_norm": 4.078341484069824,
- "learning_rate": 0.0001156043956043956,
- "loss": 1.4392,
- "step": 19200
- },
- {
- "epoch": 2.120879120879121,
- "grad_norm": 4.639175891876221,
- "learning_rate": 0.00011516483516483517,
- "loss": 1.4325,
- "step": 19300
- },
- {
- "epoch": 2.131868131868132,
- "grad_norm": 3.981069326400757,
- "learning_rate": 0.00011472527472527473,
- "loss": 1.4833,
- "step": 19400
- },
  {
  "epoch": 2.142857142857143,
- "grad_norm": 4.465806007385254,
+ "grad_norm": 5.382218837738037,
  "learning_rate": 0.00011428571428571428,
- "loss": 1.4651,
+ "loss": 1.3463,
  "step": 19500
  },
- {
- "epoch": 2.1538461538461537,
- "grad_norm": 5.009954929351807,
- "learning_rate": 0.00011384615384615384,
- "loss": 1.4562,
- "step": 19600
- },
- {
- "epoch": 2.1648351648351647,
- "grad_norm": 6.226076126098633,
- "learning_rate": 0.00011340659340659341,
- "loss": 1.4481,
- "step": 19700
- },
- {
- "epoch": 2.1758241758241756,
- "grad_norm": 5.82151985168457,
- "learning_rate": 0.00011296703296703297,
- "loss": 1.4518,
- "step": 19800
- },
- {
- "epoch": 2.186813186813187,
- "grad_norm": 4.184169292449951,
- "learning_rate": 0.00011252747252747253,
- "loss": 1.428,
- "step": 19900
- },
  {
  "epoch": 2.197802197802198,
- "grad_norm": 9.352435111999512,
+ "grad_norm": 3.35638427734375,
  "learning_rate": 0.0001120879120879121,
- "loss": 1.4296,
+ "loss": 1.3418,
  "step": 20000
  },
- {
- "epoch": 2.208791208791209,
- "grad_norm": 4.715268611907959,
- "learning_rate": 0.00011164835164835166,
- "loss": 1.4184,
- "step": 20100
- },
- {
- "epoch": 2.21978021978022,
- "grad_norm": 4.748434543609619,
- "learning_rate": 0.00011120879120879122,
- "loss": 1.4969,
- "step": 20200
- },
- {
- "epoch": 2.230769230769231,
- "grad_norm": 4.664528846740723,
- "learning_rate": 0.00011076923076923077,
- "loss": 1.4535,
- "step": 20300
- },
- {
- "epoch": 2.241758241758242,
- "grad_norm": 4.951592445373535,
- "learning_rate": 0.00011032967032967035,
- "loss": 1.4665,
- "step": 20400
- },
  {
  "epoch": 2.2527472527472527,
- "grad_norm": 4.155642032623291,
+ "grad_norm": 3.04825496673584,
  "learning_rate": 0.0001098901098901099,
- "loss": 1.4741,
+ "loss": 1.3568,
  "step": 20500
  },
- {
- "epoch": 2.2637362637362637,
- "grad_norm": 4.944486618041992,
- "learning_rate": 0.00010945054945054946,
- "loss": 1.4623,
- "step": 20600
- },
- {
- "epoch": 2.2747252747252746,
- "grad_norm": 5.376631259918213,
- "learning_rate": 0.000109010989010989,
- "loss": 1.4411,
- "step": 20700
- },
- {
- "epoch": 2.2857142857142856,
- "grad_norm": 5.888622760772705,
- "learning_rate": 0.00010857142857142856,
- "loss": 1.4515,
- "step": 20800
- },
- {
- "epoch": 2.2967032967032965,
- "grad_norm": 4.544143199920654,
- "learning_rate": 0.00010813186813186814,
- "loss": 1.4128,
- "step": 20900
- },
  {
  "epoch": 2.3076923076923075,
- "grad_norm": 5.329171657562256,
+ "grad_norm": 2.9136993885040283,
  "learning_rate": 0.0001076923076923077,
- "loss": 1.4877,
+ "loss": 1.3507,
  "step": 21000
  },
- {
- "epoch": 2.3186813186813184,
- "grad_norm": 5.327804088592529,
- "learning_rate": 0.00010725274725274725,
- "loss": 1.4718,
- "step": 21100
- },
- {
- "epoch": 2.32967032967033,
- "grad_norm": 5.668950080871582,
- "learning_rate": 0.00010681318681318682,
- "loss": 1.4485,
- "step": 21200
- },
- {
- "epoch": 2.340659340659341,
- "grad_norm": 4.657803058624268,
- "learning_rate": 0.00010637362637362638,
- "loss": 1.4781,
- "step": 21300
- },
- {
- "epoch": 2.3516483516483517,
- "grad_norm": 5.519957065582275,
- "learning_rate": 0.00010593406593406594,
- "loss": 1.4356,
- "step": 21400
- },
  {
  "epoch": 2.3626373626373627,
- "grad_norm": 4.988160133361816,
+ "grad_norm": 2.5451395511627197,
  "learning_rate": 0.0001054945054945055,
- "loss": 1.4502,
+ "loss": 1.353,
  "step": 21500
  },
- {
- "epoch": 2.3736263736263736,
- "grad_norm": 7.092361927032471,
- "learning_rate": 0.00010505494505494507,
- "loss": 1.4482,
- "step": 21600
- },
- {
- "epoch": 2.3846153846153846,
- "grad_norm": 3.946814775466919,
- "learning_rate": 0.00010461538461538463,
- "loss": 1.45,
- "step": 21700
- },
- {
- "epoch": 2.3956043956043955,
- "grad_norm": 7.119655132293701,
- "learning_rate": 0.00010417582417582418,
- "loss": 1.4431,
- "step": 21800
- },
- {
- "epoch": 2.4065934065934065,
- "grad_norm": 4.780879974365234,
- "learning_rate": 0.00010373626373626376,
- "loss": 1.4483,
- "step": 21900
- },
  {
  "epoch": 2.4175824175824174,
- "grad_norm": 7.585813045501709,
+ "grad_norm": 2.779684066772461,
  "learning_rate": 0.00010329670329670331,
- "loss": 1.4655,
+ "loss": 1.3456,
  "step": 22000
  },
- {
- "epoch": 2.4285714285714284,
- "grad_norm": 3.923065662384033,
- "learning_rate": 0.00010285714285714286,
- "loss": 1.4625,
- "step": 22100
- },
- {
- "epoch": 2.4395604395604398,
- "grad_norm": 4.095515727996826,
- "learning_rate": 0.00010241758241758242,
- "loss": 1.4558,
- "step": 22200
- },
- {
- "epoch": 2.4505494505494507,
- "grad_norm": 3.7886769771575928,
- "learning_rate": 0.00010197802197802197,
- "loss": 1.4966,
- "step": 22300
- },
- {
- "epoch": 2.4615384615384617,
- "grad_norm": 5.163322925567627,
- "learning_rate": 0.00010153846153846153,
- "loss": 1.4471,
- "step": 22400
- },
  {
  "epoch": 2.4725274725274726,
- "grad_norm": 3.9905765056610107,
+ "grad_norm": 3.8211276531219482,
  "learning_rate": 0.0001010989010989011,
- "loss": 1.4346,
+ "loss": 1.3519,
  "step": 22500
  },
- {
- "epoch": 2.4835164835164836,
- "grad_norm": 6.910905838012695,
- "learning_rate": 0.00010065934065934066,
- "loss": 1.4391,
- "step": 22600
- },
- {
- "epoch": 2.4945054945054945,
- "grad_norm": 4.383417129516602,
- "learning_rate": 0.00010021978021978022,
- "loss": 1.4457,
- "step": 22700
- },
- {
- "epoch": 2.5054945054945055,
- "grad_norm": 4.188605308532715,
- "learning_rate": 9.978021978021979e-05,
- "loss": 1.4623,
- "step": 22800
- },
- {
- "epoch": 2.5164835164835164,
- "grad_norm": 3.304708242416382,
- "learning_rate": 9.934065934065935e-05,
- "loss": 1.4539,
- "step": 22900
- },
  {
  "epoch": 2.5274725274725274,
- "grad_norm": 6.389969348907471,
+ "grad_norm": 2.877223253250122,
  "learning_rate": 9.89010989010989e-05,
- "loss": 1.4642,
+ "loss": 1.3506,
  "step": 23000
  },
- {
- "epoch": 2.5384615384615383,
- "grad_norm": 4.69569730758667,
- "learning_rate": 9.846153846153848e-05,
- "loss": 1.4674,
- "step": 23100
- },
- {
- "epoch": 2.5494505494505493,
- "grad_norm": 5.309410572052002,
- "learning_rate": 9.802197802197802e-05,
- "loss": 1.4758,
- "step": 23200
- },
- {
- "epoch": 2.5604395604395602,
- "grad_norm": 5.060328483581543,
- "learning_rate": 9.758241758241758e-05,
- "loss": 1.4705,
- "step": 23300
- },
- {
- "epoch": 2.571428571428571,
- "grad_norm": 5.109841346740723,
- "learning_rate": 9.714285714285715e-05,
- "loss": 1.4226,
- "step": 23400
- },
  {
  "epoch": 2.5824175824175826,
- "grad_norm": 4.877140522003174,
+ "grad_norm": 2.53200101852417,
  "learning_rate": 9.670329670329671e-05,
- "loss": 1.466,
+ "loss": 1.3564,
  "step": 23500
  },
- {
- "epoch": 2.5934065934065935,
- "grad_norm": 5.292356967926025,
- "learning_rate": 9.626373626373627e-05,
- "loss": 1.4619,
- "step": 23600
- },
- {
- "epoch": 2.6043956043956045,
- "grad_norm": 4.887085914611816,
- "learning_rate": 9.582417582417584e-05,
- "loss": 1.4379,
- "step": 23700
- },
- {
- "epoch": 2.6153846153846154,
- "grad_norm": 4.668257713317871,
- "learning_rate": 9.53846153846154e-05,
- "loss": 1.4249,
- "step": 23800
- },
- {
- "epoch": 2.6263736263736264,
- "grad_norm": 6.5269293785095215,
- "learning_rate": 9.494505494505494e-05,
- "loss": 1.447,
- "step": 23900
- },
  {
  "epoch": 2.6373626373626373,
- "grad_norm": 4.972898483276367,
+ "grad_norm": 3.249028205871582,
  "learning_rate": 9.450549450549451e-05,
- "loss": 1.4645,
+ "loss": 1.3393,
  "step": 24000
  },
- {
- "epoch": 2.6483516483516483,
- "grad_norm": 5.63477897644043,
- "learning_rate": 9.406593406593407e-05,
- "loss": 1.4754,
- "step": 24100
- },
- {
- "epoch": 2.659340659340659,
- "grad_norm": 4.150704383850098,
- "learning_rate": 9.362637362637363e-05,
- "loss": 1.4467,
- "step": 24200
- },
- {
- "epoch": 2.67032967032967,
- "grad_norm": 6.260478973388672,
- "learning_rate": 9.318681318681319e-05,
- "loss": 1.4454,
- "step": 24300
- },
- {
- "epoch": 2.6813186813186816,
- "grad_norm": 6.3307929039001465,
- "learning_rate": 9.274725274725276e-05,
- "loss": 1.4387,
- "step": 24400
- },
  {
  "epoch": 2.6923076923076925,
- "grad_norm": 7.524434566497803,
+ "grad_norm": 2.819939613342285,
  "learning_rate": 9.230769230769232e-05,
- "loss": 1.4203,
+ "loss": 1.3413,
  "step": 24500
  },
- {
- "epoch": 2.7032967032967035,
- "grad_norm": 5.355330467224121,
- "learning_rate": 9.186813186813187e-05,
- "loss": 1.4282,
- "step": 24600
- },
- {
- "epoch": 2.7142857142857144,
- "grad_norm": 4.880154132843018,
- "learning_rate": 9.142857142857143e-05,
- "loss": 1.4258,
- "step": 24700
- },
- {
- "epoch": 2.7252747252747254,
- "grad_norm": 6.773056983947754,
- "learning_rate": 9.098901098901099e-05,
- "loss": 1.4809,
- "step": 24800
- },
- {
- "epoch": 2.7362637362637363,
- "grad_norm": 4.716909408569336,
- "learning_rate": 9.054945054945055e-05,
- "loss": 1.4571,
- "step": 24900
- },
  {
  "epoch": 2.7472527472527473,
- "grad_norm": 8.557913780212402,
+ "grad_norm": 3.4391469955444336,
  "learning_rate": 9.010989010989012e-05,
- "loss": 1.486,
+ "loss": 1.3517,
  "step": 25000
  },
- {
- "epoch": 2.758241758241758,
- "grad_norm": 7.815650939941406,
- "learning_rate": 8.967032967032968e-05,
- "loss": 1.4285,
- "step": 25100
- },
- {
- "epoch": 2.769230769230769,
- "grad_norm": 3.7082254886627197,
- "learning_rate": 8.923076923076924e-05,
- "loss": 1.4619,
- "step": 25200
- },
- {
- "epoch": 2.78021978021978,
- "grad_norm": 5.004034996032715,
- "learning_rate": 8.87912087912088e-05,
- "loss": 1.4389,
- "step": 25300
- },
- {
- "epoch": 2.791208791208791,
- "grad_norm": 4.239031791687012,
- "learning_rate": 8.835164835164835e-05,
- "loss": 1.4405,
- "step": 25400
- },
  {
  "epoch": 2.802197802197802,
- "grad_norm": 7.5476393699646,
+ "grad_norm": 2.9319329261779785,
  "learning_rate": 8.791208791208791e-05,
- "loss": 1.4206,
+ "loss": 1.3362,
  "step": 25500
  },
- {
- "epoch": 2.813186813186813,
- "grad_norm": 2.9742252826690674,
- "learning_rate": 8.747252747252748e-05,
- "loss": 1.4213,
- "step": 25600
- },
- {
- "epoch": 2.824175824175824,
- "grad_norm": 4.981147289276123,
- "learning_rate": 8.703296703296704e-05,
- "loss": 1.4565,
- "step": 25700
- },
- {
- "epoch": 2.8351648351648353,
- "grad_norm": 4.742037296295166,
- "learning_rate": 8.65934065934066e-05,
- "loss": 1.44,
- "step": 25800
- },
- {
- "epoch": 2.8461538461538463,
- "grad_norm": 5.7680253982543945,
- "learning_rate": 8.615384615384617e-05,
- "loss": 1.4416,
- "step": 25900
- },
  {
  "epoch": 2.857142857142857,
- "grad_norm": 5.002392768859863,
+ "grad_norm": 2.257335901260376,
  "learning_rate": 8.571428571428571e-05,
- "loss": 1.4247,
+ "loss": 1.3402,
  "step": 26000
  },
- {
- "epoch": 2.868131868131868,
- "grad_norm": 6.970800399780273,
- "learning_rate": 8.527472527472527e-05,
- "loss": 1.4754,
- "step": 26100
- },
- {
- "epoch": 2.879120879120879,
- "grad_norm": 4.084986686706543,
- "learning_rate": 8.483516483516484e-05,
- "loss": 1.4369,
- "step": 26200
- },
- {
- "epoch": 2.89010989010989,
- "grad_norm": 5.708220481872559,
- "learning_rate": 8.43956043956044e-05,
- "loss": 1.4711,
- "step": 26300
- },
- {
- "epoch": 2.901098901098901,
- "grad_norm": 4.949507713317871,
- "learning_rate": 8.395604395604396e-05,
- "loss": 1.4651,
- "step": 26400
- },
  {
  "epoch": 2.912087912087912,
- "grad_norm": 4.25984001159668,
+ "grad_norm": 2.6121294498443604,
  "learning_rate": 8.351648351648353e-05,
- "loss": 1.4692,
+ "loss": 1.3615,
  "step": 26500
  },
- {
- "epoch": 2.9230769230769234,
- "grad_norm": 3.862776756286621,
- "learning_rate": 8.307692307692309e-05,
- "loss": 1.437,
- "step": 26600
- },
- {
- "epoch": 2.9340659340659343,
- "grad_norm": 4.923271656036377,
- "learning_rate": 8.263736263736263e-05,
- "loss": 1.4416,
- "step": 26700
- },
- {
- "epoch": 2.9450549450549453,
- "grad_norm": 4.893067359924316,
- "learning_rate": 8.21978021978022e-05,
- "loss": 1.4585,
- "step": 26800
- },
- {
- "epoch": 2.956043956043956,
- "grad_norm": 4.829097747802734,
- "learning_rate": 8.175824175824176e-05,
- "loss": 1.4411,
- "step": 26900
- },
  {
  "epoch": 2.967032967032967,
- "grad_norm": 5.341473579406738,
+ "grad_norm": 2.716127395629883,
  "learning_rate": 8.131868131868132e-05,
- "loss": 1.4662,
+ "loss": 1.3418,
  "step": 27000
  },
- {
- "epoch": 2.978021978021978,
- "grad_norm": 5.461435317993164,
- "learning_rate": 8.087912087912089e-05,
- "loss": 1.4576,
- "step": 27100
- },
- {
- "epoch": 2.989010989010989,
- "grad_norm": 4.598509311676025,
- "learning_rate": 8.043956043956045e-05,
- "loss": 1.4681,
- "step": 27200
- },
- {
- "epoch": 3.0,
- "grad_norm": 5.4833149909973145,
- "learning_rate": 8e-05,
- "loss": 1.4727,
- "step": 27300
- },
- {
- "epoch": 3.010989010989011,
- "grad_norm": 4.663654327392578,
- "learning_rate": 7.956043956043956e-05,
- "loss": 1.3761,
- "step": 27400
- },
  {
  "epoch": 3.021978021978022,
- "grad_norm": 6.357902526855469,
+ "grad_norm": 2.227154493331909,
  "learning_rate": 7.912087912087912e-05,
- "loss": 1.3546,
+ "loss": 1.3199,
  "step": 27500
  },
- {
- "epoch": 3.032967032967033,
- "grad_norm": 4.271589279174805,
- "learning_rate": 7.868131868131868e-05,
- "loss": 1.3362,
- "step": 27600
- },
- {
- "epoch": 3.043956043956044,
- "grad_norm": 6.143280506134033,
- "learning_rate": 7.824175824175824e-05,
- "loss": 1.3565,
- "step": 27700
- },
- {
- "epoch": 3.0549450549450547,
- "grad_norm": 8.49393081665039,
- "learning_rate": 7.780219780219781e-05,
- "loss": 1.3434,
- "step": 27800
- },
- {
- "epoch": 3.065934065934066,
- "grad_norm": 5.024664878845215,
- "learning_rate": 7.736263736263737e-05,
- "loss": 1.3385,
- "step": 27900
- },
  {
  "epoch": 3.076923076923077,
- "grad_norm": 6.776920795440674,
+ "grad_norm": 3.0606658458709717,
  "learning_rate": 7.692307692307693e-05,
- "loss": 1.3671,
+ "loss": 1.2553,
  "step": 28000
  },
- {
- "epoch": 3.087912087912088,
- "grad_norm": 5.72268009185791,
- "learning_rate": 7.648351648351648e-05,
- "loss": 1.3673,
- "step": 28100
- },
- {
- "epoch": 3.098901098901099,
- "grad_norm": 5.348138332366943,
- "learning_rate": 7.604395604395604e-05,
- "loss": 1.3505,
- "step": 28200
- },
- {
- "epoch": 3.10989010989011,
- "grad_norm": 4.440935134887695,
- "learning_rate": 7.56043956043956e-05,
- "loss": 1.3583,
- "step": 28300
- },
- {
- "epoch": 3.120879120879121,
- "grad_norm": 4.01630163192749,
- "learning_rate": 7.516483516483517e-05,
- "loss": 1.3524,
- "step": 28400
- },
  {
  "epoch": 3.131868131868132,
- "grad_norm": 6.5499043464660645,
+ "grad_norm": 2.8261947631835938,
  "learning_rate": 7.472527472527473e-05,
- "loss": 1.3548,
+ "loss": 1.2688,
  "step": 28500
  },
- {
- "epoch": 3.142857142857143,
- "grad_norm": 5.873259544372559,
- "learning_rate": 7.428571428571429e-05,
- "loss": 1.3408,
- "step": 28600
- },
- {
- "epoch": 3.1538461538461537,
- "grad_norm": 4.558487892150879,
- "learning_rate": 7.384615384615386e-05,
- "loss": 1.3017,
- "step": 28700
- },
- {
- "epoch": 3.1648351648351647,
- "grad_norm": 7.020089149475098,
- "learning_rate": 7.34065934065934e-05,
- "loss": 1.3585,
- "step": 28800
- },
- {
- "epoch": 3.1758241758241756,
- "grad_norm": 4.015739917755127,
- "learning_rate": 7.296703296703296e-05,
- "loss": 1.3393,
- "step": 28900
- },
  {
  "epoch": 3.186813186813187,
- "grad_norm": 4.7826972007751465,
+ "grad_norm": 3.738393783569336,
  "learning_rate": 7.252747252747253e-05,
- "loss": 1.371,
+ "loss": 1.2513,
  "step": 29000
  },
- {
- "epoch": 3.197802197802198,
- "grad_norm": 9.16402816772461,
- "learning_rate": 7.208791208791209e-05,
- "loss": 1.341,
- "step": 29100
- },
- {
- "epoch": 3.208791208791209,
- "grad_norm": 5.517131328582764,
- "learning_rate": 7.164835164835165e-05,
- "loss": 1.3631,
- "step": 29200
- },
- {
- "epoch": 3.21978021978022,
- "grad_norm": 9.678403854370117,
- "learning_rate": 7.120879120879122e-05,
- "loss": 1.3739,
- "step": 29300
- },
- {
- "epoch": 3.230769230769231,
- "grad_norm": 5.7049880027771,
- "learning_rate": 7.076923076923078e-05,
- "loss": 1.3897,
- "step": 29400
- },
  {
  "epoch": 3.241758241758242,
- "grad_norm": 6.8695292472839355,
+ "grad_norm": 2.5947108268737793,
  "learning_rate": 7.032967032967034e-05,
- "loss": 1.3764,
+ "loss": 1.2757,
  "step": 29500
  },
- {
- "epoch": 3.2527472527472527,
- "grad_norm": 7.1398115158081055,
- "learning_rate": 6.98901098901099e-05,
- "loss": 1.3695,
- "step": 29600
- },
- {
- "epoch": 3.2637362637362637,
- "grad_norm": 5.985105991363525,
- "learning_rate": 6.945054945054945e-05,
- "loss": 1.3642,
- "step": 29700
- },
- {
- "epoch": 3.2747252747252746,
- "grad_norm": 6.806448459625244,
- "learning_rate": 6.901098901098901e-05,
- "loss": 1.3625,
- "step": 29800
- },
- {
- "epoch": 3.2857142857142856,
- "grad_norm": 5.546496868133545,
- "learning_rate": 6.857142857142858e-05,
- "loss": 1.3715,
- "step": 29900
- },
  {
  "epoch": 3.2967032967032965,
- "grad_norm": 5.092505931854248,
+ "grad_norm": 2.707387924194336,
  "learning_rate": 6.813186813186814e-05,
- "loss": 1.3758,
+ "loss": 1.274,
  "step": 30000
  },
- {
- "epoch": 3.3076923076923075,
- "grad_norm": 5.186868190765381,
- "learning_rate": 6.76923076923077e-05,
- "loss": 1.3715,
- "step": 30100
- },
- {
- "epoch": 3.3186813186813184,
- "grad_norm": 5.597959995269775,
- "learning_rate": 6.725274725274725e-05,
- "loss": 1.357,
- "step": 30200
- },
- {
- "epoch": 3.32967032967033,
- "grad_norm": 5.903227806091309,
- "learning_rate": 6.681318681318681e-05,
- "loss": 1.3479,
- "step": 30300
- },
- {
- "epoch": 3.340659340659341,
- "grad_norm": 5.141580581665039,
- "learning_rate": 6.637362637362637e-05,
- "loss": 1.3734,
- "step": 30400
- },
  {
  "epoch": 3.3516483516483517,
- "grad_norm": 9.274611473083496,
+ "grad_norm": 3.4624383449554443,
  "learning_rate": 6.593406593406594e-05,
- "loss": 1.3423,
+ "loss": 1.2702,
  "step": 30500
  },
- {
- "epoch": 3.3626373626373627,
- "grad_norm": 5.92864465713501,
- "learning_rate": 6.54945054945055e-05,
- "loss": 1.358,
- "step": 30600
- },
- {
- "epoch": 3.3736263736263736,
- "grad_norm": 5.521232604980469,
- "learning_rate": 6.505494505494506e-05,
- "loss": 1.3117,
- "step": 30700
- },
- {
- "epoch": 3.3846153846153846,
- "grad_norm": 6.751832962036133,
- "learning_rate": 6.461538461538462e-05,
- "loss": 1.3345,
- "step": 30800
- },
- {
- "epoch": 3.3956043956043955,
- "grad_norm": 7.770620346069336,
- "learning_rate": 6.417582417582419e-05,
- "loss": 1.3192,
- "step": 30900
- },
  {
  "epoch": 3.4065934065934065,
- "grad_norm": 4.924952983856201,
+ "grad_norm": 3.7795355319976807,
  "learning_rate": 6.373626373626373e-05,
- "loss": 1.3553,
+ "loss": 1.2409,
  "step": 31000
  },
- {
- "epoch": 3.4175824175824174,
- "grad_norm": 6.34819221496582,
- "learning_rate": 6.32967032967033e-05,
- "loss": 1.3726,
- "step": 31100
- },
- {
- "epoch": 3.4285714285714284,
- "grad_norm": 7.109829425811768,
- "learning_rate": 6.285714285714286e-05,
- "loss": 1.3753,
- "step": 31200
- },
- {
- "epoch": 3.4395604395604398,
- "grad_norm": 6.445315361022949,
- "learning_rate": 6.241758241758242e-05,
- "loss": 1.3552,
- "step": 31300
- },
- {
- "epoch": 3.4505494505494507,
- "grad_norm": 4.920654296875,
- "learning_rate": 6.197802197802198e-05,
- "loss": 1.356,
- "step": 31400
- },
  {
  "epoch": 3.4615384615384617,
- "grad_norm": 3.755321741104126,
+ "grad_norm": 3.521811008453369,
  "learning_rate": 6.153846153846155e-05,
- "loss": 1.3396,
+ "loss": 1.267,
  "step": 31500
  },
- {
- "epoch": 3.4725274725274726,
- "grad_norm": 7.977355480194092,
- "learning_rate": 6.10989010989011e-05,
- "loss": 1.3677,
- "step": 31600
- },
- {
- "epoch": 3.4835164835164836,
- "grad_norm": 7.270201206207275,
- "learning_rate": 6.065934065934066e-05,
- "loss": 1.3782,
- "step": 31700
- },
- {
- "epoch": 3.4945054945054945,
- "grad_norm": 4.333983898162842,
- "learning_rate": 6.021978021978022e-05,
- "loss": 1.3477,
- "step": 31800
- },
- {
- "epoch": 3.5054945054945055,
- "grad_norm": 5.169381141662598,
- "learning_rate": 5.978021978021978e-05,
- "loss": 1.3181,
- "step": 31900
- },
  {
  "epoch": 3.5164835164835164,
- "grad_norm": 6.523856163024902,
+ "grad_norm": 2.7402503490448,
  "learning_rate": 5.9340659340659345e-05,
- "loss": 1.3355,
+ "loss": 1.2641,
  "step": 32000
  },
- {
- "epoch": 3.5274725274725274,
- "grad_norm": 5.986196041107178,
- "learning_rate": 5.89010989010989e-05,
- "loss": 1.3586,
- "step": 32100
- },
- {
- "epoch": 3.5384615384615383,
- "grad_norm": 5.404372215270996,
- "learning_rate": 5.846153846153847e-05,
- "loss": 1.3575,
- "step": 32200
- },
- {
- "epoch": 3.5494505494505493,
- "grad_norm": 4.96329927444458,
- "learning_rate": 5.802197802197803e-05,
- "loss": 1.3882,
- "step": 32300
- },
- {
- "epoch": 3.5604395604395602,
- "grad_norm": 8.159017562866211,
- "learning_rate": 5.7582417582417584e-05,
- "loss": 1.3239,
- "step": 32400
- },
  {
  "epoch": 3.571428571428571,
- "grad_norm": 3.772956371307373,
+ "grad_norm": 2.937671661376953,
  "learning_rate": 5.714285714285714e-05,
- "loss": 1.3597,
+ "loss": 1.2626,
  "step": 32500
  },
- {
- "epoch": 3.5824175824175826,
- "grad_norm": 4.539822578430176,
- "learning_rate": 5.6703296703296706e-05,
- "loss": 1.3353,
- "step": 32600
- },
- {
- "epoch": 3.5934065934065935,
- "grad_norm": 7.850346088409424,
- "learning_rate": 5.6263736263736264e-05,
- "loss": 1.3523,
- "step": 32700
- },
- {
- "epoch": 3.6043956043956045,
- "grad_norm": 6.533629417419434,
- "learning_rate": 5.582417582417583e-05,
- "loss": 1.3777,
- "step": 32800
- },
- {
- "epoch": 3.6153846153846154,
- "grad_norm": 4.991445064544678,
- "learning_rate": 5.538461538461539e-05,
- "loss": 1.358,
- "step": 32900
- },
  {
  "epoch": 3.6263736263736264,
- "grad_norm": 4.861871242523193,
+ "grad_norm": 2.886845350265503,
  "learning_rate": 5.494505494505495e-05,
- "loss": 1.3785,
+ "loss": 1.269,
  "step": 33000
  },
- {
- "epoch": 3.6373626373626373,
- "grad_norm": 15.688408851623535,
- "learning_rate": 5.45054945054945e-05,
- "loss": 1.3492,
- "step": 33100
- },
- {
- "epoch": 3.6483516483516483,
- "grad_norm": 5.074567794799805,
- "learning_rate": 5.406593406593407e-05,
- "loss": 1.3764,
- "step": 33200
- },
- {
- "epoch": 3.659340659340659,
- "grad_norm": 6.356872081756592,
- "learning_rate": 5.3626373626373626e-05,
- "loss": 1.3701,
- "step": 33300
- },
- {
- "epoch": 3.67032967032967,
- "grad_norm": 4.222293376922607,
- "learning_rate": 5.318681318681319e-05,
- "loss": 1.3443,
- "step": 33400
- },
  {
  "epoch": 3.6813186813186816,
- "grad_norm": 6.198267459869385,
+ "grad_norm": 3.9152579307556152,
  "learning_rate": 5.274725274725275e-05,
- "loss": 1.3527,
+ "loss": 1.2659,
  "step": 33500
  },
- {
- "epoch": 3.6923076923076925,
- "grad_norm": 5.048154354095459,
- "learning_rate": 5.230769230769231e-05,
- "loss": 1.318,
- "step": 33600
- },
- {
- "epoch": 3.7032967032967035,
- "grad_norm": 4.227085590362549,
- "learning_rate": 5.186813186813188e-05,
- "loss": 1.3581,
- "step": 33700
- },
- {
- "epoch": 3.7142857142857144,
- "grad_norm": 9.587520599365234,
- "learning_rate": 5.142857142857143e-05,
- "loss": 1.3228,
- "step": 33800
- },
- {
- "epoch": 3.7252747252747254,
- "grad_norm": 6.039963722229004,
- "learning_rate": 5.098901098901099e-05,
- "loss": 1.349,
- "step": 33900
- },
  {
  "epoch": 3.7362637362637363,
- "grad_norm": 3.8198747634887695,
+ "grad_norm": 2.7805376052856445,
  "learning_rate": 5.054945054945055e-05,
- "loss": 1.3624,
+ "loss": 1.2523,
  "step": 34000
  },
- {
- "epoch": 3.7472527472527473,
- "grad_norm": 10.933670043945312,
- "learning_rate": 5.010989010989011e-05,
- "loss": 1.3313,
- "step": 34100
- },
- {
- "epoch": 3.758241758241758,
- "grad_norm": 6.5416364669799805,
- "learning_rate": 4.9670329670329674e-05,
- "loss": 1.3401,
- "step": 34200
- },
- {
- "epoch": 3.769230769230769,
- "grad_norm": 4.092519283294678,
- "learning_rate": 4.923076923076924e-05,
- "loss": 1.349,
- "step": 34300
- },
- {
- "epoch": 3.78021978021978,
- "grad_norm": 6.520788669586182,
- "learning_rate": 4.879120879120879e-05,
- "loss": 1.3346,
- "step": 34400
- },
  {
  "epoch": 3.791208791208791,
- "grad_norm": 5.3541460037231445,
+ "grad_norm": 3.7305562496185303,
  "learning_rate": 4.8351648351648355e-05,
- "loss": 1.3657,
+ "loss": 1.2528,
  "step": 34500
  },
- {
- "epoch": 3.802197802197802,
- "grad_norm": 4.871542453765869,
- "learning_rate": 4.791208791208792e-05,
- "loss": 1.3517,
- "step": 34600
- },
- {
- "epoch": 3.813186813186813,
- "grad_norm": 6.653789520263672,
- "learning_rate": 4.747252747252747e-05,
- "loss": 1.356,
- "step": 34700
- },
- {
- "epoch": 3.824175824175824,
- "grad_norm": 7.272556781768799,
- "learning_rate": 4.7032967032967035e-05,
- "loss": 1.3392,
- "step": 34800
- },
- {
- "epoch": 3.8351648351648353,
- "grad_norm": 9.819136619567871,
- "learning_rate": 4.6593406593406593e-05,
- "loss": 1.3772,
- "step": 34900
- },
  {
  "epoch": 3.8461538461538463,
- "grad_norm": 5.430606365203857,
+ "grad_norm": 3.0859031677246094,
  "learning_rate": 4.615384615384616e-05,
- "loss": 1.3893,
+ "loss": 1.273,
  "step": 35000
  },
- {
- "epoch": 3.857142857142857,
- "grad_norm": 5.265222549438477,
- "learning_rate": 4.5714285714285716e-05,
- "loss": 1.3609,
- "step": 35100
- },
- {
- "epoch": 3.868131868131868,
- "grad_norm": 6.421568870544434,
- "learning_rate": 4.5274725274725274e-05,
- "loss": 1.3848,
- "step": 35200
- },
- {
- "epoch": 3.879120879120879,
- "grad_norm": 5.183870315551758,
- "learning_rate": 4.483516483516484e-05,
- "loss": 1.3554,
- "step": 35300
- },
- {
- "epoch": 3.89010989010989,
- "grad_norm": 6.2706170082092285,
- "learning_rate": 4.43956043956044e-05,
- "loss": 1.3603,
- "step": 35400
- },
  {
  "epoch": 3.901098901098901,
- "grad_norm": 5.7601518630981445,
+ "grad_norm": 4.31486177444458,
  "learning_rate": 4.3956043956043955e-05,
- "loss": 1.341,
+ "loss": 1.2739,
  "step": 35500
  },
- {
- "epoch": 3.912087912087912,
- "grad_norm": 6.289906024932861,
- "learning_rate": 4.351648351648352e-05,
- "loss": 1.3188,
- "step": 35600
- },
- {
- "epoch": 3.9230769230769234,
- "grad_norm": 8.751489639282227,
- "learning_rate": 4.3076923076923084e-05,
- "loss": 1.3744,
- "step": 35700
- },
- {
- "epoch": 3.9340659340659343,
- "grad_norm": 8.174962043762207,
- "learning_rate": 4.2637362637362635e-05,
- "loss": 1.3422,
- "step": 35800
- },
- {
- "epoch": 3.9450549450549453,
- "grad_norm": 5.522929668426514,
- "learning_rate": 4.21978021978022e-05,
- "loss": 1.3361,
- "step": 35900
- },
  {
  "epoch": 3.956043956043956,
- "grad_norm": 4.805854797363281,
+ "grad_norm": 2.46838116645813,
  "learning_rate": 4.1758241758241765e-05,
- "loss": 1.3541,
+ "loss": 1.2555,
  "step": 36000
  },
- {
- "epoch": 3.967032967032967,
- "grad_norm": 5.191071510314941,
- "learning_rate": 4.1318681318681316e-05,
- "loss": 1.3534,
- "step": 36100
- },
- {
- "epoch": 3.978021978021978,
- "grad_norm": 5.726668357849121,
- "learning_rate": 4.087912087912088e-05,
- "loss": 1.3106,
- "step": 36200
- },
- {
- "epoch": 3.989010989010989,
- "grad_norm": 3.8736844062805176,
- "learning_rate": 4.0439560439560445e-05,
- "loss": 1.3477,
- "step": 36300
- },
- {
- "epoch": 4.0,
- "grad_norm": 5.525221824645996,
- "learning_rate": 4e-05,
- "loss": 1.3572,
- "step": 36400
- },
  {
  "epoch": 4.010989010989011,
- "grad_norm": 3.638927698135376,
+ "grad_norm": 3.3113248348236084,
  "learning_rate": 3.956043956043956e-05,
- "loss": 1.2686,
+ "loss": 1.2388,
  "step": 36500
  },
- {
- "epoch": 4.021978021978022,
- "grad_norm": 7.469931125640869,
- "learning_rate": 3.912087912087912e-05,
- "loss": 1.2631,
- "step": 36600
- },
- {
- "epoch": 4.032967032967033,
- "grad_norm": 5.60840368270874,
- "learning_rate": 3.8681318681318684e-05,
- "loss": 1.2658,
- "step": 36700
- },
- {
- "epoch": 4.043956043956044,
- "grad_norm": 5.429786205291748,
- "learning_rate": 3.824175824175824e-05,
- "loss": 1.2783,
- "step": 36800
- },
- {
- "epoch": 4.054945054945055,
- "grad_norm": 6.977511882781982,
- "learning_rate": 3.78021978021978e-05,
- "loss": 1.2748,
- "step": 36900
- },
  {
  "epoch": 4.065934065934066,
- "grad_norm": 5.217138767242432,
+ "grad_norm": 5.830134868621826,
  "learning_rate": 3.7362637362637365e-05,
- "loss": 1.2688,
+ "loss": 1.1952,
  "step": 37000
  },
- {
- "epoch": 4.076923076923077,
- "grad_norm": 5.720318794250488,
- "learning_rate": 3.692307692307693e-05,
- "loss": 1.3083,
- "step": 37100
- },
- {
- "epoch": 4.087912087912088,
- "grad_norm": 7.204798698425293,
- "learning_rate": 3.648351648351648e-05,
- "loss": 1.2572,
- "step": 37200
- },
- {
- "epoch": 4.0989010989010985,
- "grad_norm": 5.261708736419678,
- "learning_rate": 3.6043956043956045e-05,
- "loss": 1.2744,
- "step": 37300
- },
- {
- "epoch": 4.1098901098901095,
- "grad_norm": 3.5920863151550293,
- "learning_rate": 3.560439560439561e-05,
- "loss": 1.2563,
- "step": 37400
- },
  {
  "epoch": 4.1208791208791204,
- "grad_norm": 6.172001361846924,
+ "grad_norm": 2.97209095954895,
  "learning_rate": 3.516483516483517e-05,
- "loss": 1.2532,
+ "loss": 1.1902,
  "step": 37500
  },
- {
- "epoch": 4.131868131868132,
- "grad_norm": 6.257133483886719,
- "learning_rate": 3.4725274725274726e-05,
- "loss": 1.2462,
- "step": 37600
- },
- {
- "epoch": 4.142857142857143,
- "grad_norm": 7.922190189361572,
- "learning_rate": 3.428571428571429e-05,
- "loss": 1.2428,
- "step": 37700
- },
- {
- "epoch": 4.153846153846154,
- "grad_norm": 6.363681316375732,
- "learning_rate": 3.384615384615385e-05,
- "loss": 1.2898,
- "step": 37800
- },
- {
- "epoch": 4.164835164835165,
- "grad_norm": 4.985836505889893,
- "learning_rate": 3.3406593406593406e-05,
- "loss": 1.2577,
- "step": 37900
- },
  {
  "epoch": 4.175824175824176,
- "grad_norm": 8.265104293823242,
+ "grad_norm": 3.9511055946350098,
  "learning_rate": 3.296703296703297e-05,
- "loss": 1.2669,
+ "loss": 1.1837,
  "step": 38000
  },
- {
- "epoch": 4.186813186813187,
- "grad_norm": 9.412073135375977,
- "learning_rate": 3.252747252747253e-05,
- "loss": 1.253,
- "step": 38100
- },
- {
- "epoch": 4.197802197802198,
- "grad_norm": 6.068290710449219,
- "learning_rate": 3.2087912087912094e-05,
- "loss": 1.2902,
- "step": 38200
- },
- {
- "epoch": 4.208791208791209,
- "grad_norm": 6.8105010986328125,
- "learning_rate": 3.164835164835165e-05,
- "loss": 1.2962,
- "step": 38300
- },
- {
- "epoch": 4.21978021978022,
- "grad_norm": 6.277568340301514,
- "learning_rate": 3.120879120879121e-05,
- "loss": 1.2482,
- "step": 38400
- },
  {
  "epoch": 4.230769230769231,
- "grad_norm": 8.057587623596191,
+ "grad_norm": 6.446171760559082,
  "learning_rate": 3.0769230769230774e-05,
- "loss": 1.2523,
+ "loss": 1.1888,
  "step": 38500
  },
- {
- "epoch": 4.241758241758242,
- "grad_norm": 10.061241149902344,
- "learning_rate": 3.032967032967033e-05,
- "loss": 1.2396,
- "step": 38600
- },
- {
- "epoch": 4.252747252747253,
- "grad_norm": 5.88912296295166,
- "learning_rate": 2.989010989010989e-05,
- "loss": 1.2677,
- "step": 38700
- },
- {
- "epoch": 4.263736263736264,
- "grad_norm": 7.349645614624023,
- "learning_rate": 2.945054945054945e-05,
- "loss": 1.2397,
- "step": 38800
- },
- {
- "epoch": 4.274725274725275,
- "grad_norm": 7.222245216369629,
- "learning_rate": 2.9010989010989016e-05,
- "loss": 1.266,
- "step": 38900
- },
  {
  "epoch": 4.285714285714286,
- "grad_norm": 10.49557113647461,
+ "grad_norm": 4.0232744216918945,
  "learning_rate": 2.857142857142857e-05,
- "loss": 1.2642,
+ "loss": 1.1752,
  "step": 39000
  },
- {
- "epoch": 4.2967032967032965,
- "grad_norm": 5.563273906707764,
- "learning_rate": 2.8131868131868132e-05,
- "loss": 1.2662,
- "step": 39100
- },
- {
- "epoch": 4.3076923076923075,
- "grad_norm": 5.392216205596924,
- "learning_rate": 2.7692307692307694e-05,
- "loss": 1.2758,
- "step": 39200
- },
- {
- "epoch": 4.318681318681318,
- "grad_norm": 6.449277400970459,
- "learning_rate": 2.725274725274725e-05,
- "loss": 1.2747,
- "step": 39300
- },
- {
- "epoch": 4.329670329670329,
- "grad_norm": 8.933030128479004,
- "learning_rate": 2.6813186813186813e-05,
- "loss": 1.253,
- "step": 39400
- },
  {
  "epoch": 4.34065934065934,
- "grad_norm": 5.619176387786865,
+ "grad_norm": 4.632090091705322,
  "learning_rate": 2.6373626373626374e-05,
- "loss": 1.3107,
+ "loss": 1.1964,
  "step": 39500
  },
- {
- "epoch": 4.351648351648351,
- "grad_norm": 5.006295204162598,
- "learning_rate": 2.593406593406594e-05,
- "loss": 1.2288,
- "step": 39600
- },
- {
- "epoch": 4.362637362637362,
- "grad_norm": 7.589299201965332,
- "learning_rate": 2.5494505494505493e-05,
- "loss": 1.2758,
- "step": 39700
- },
- {
- "epoch": 4.373626373626374,
- "grad_norm": 7.450462818145752,
- "learning_rate": 2.5054945054945055e-05,
- "loss": 1.2594,
- "step": 39800
- },
- {
- "epoch": 4.384615384615385,
- "grad_norm": 6.71153450012207,
- "learning_rate": 2.461538461538462e-05,
- "loss": 1.2853,
- "step": 39900
- },
  {
  "epoch": 4.395604395604396,
- "grad_norm": 6.485320568084717,
+ "grad_norm": 3.9423835277557373,
  "learning_rate": 2.4175824175824177e-05,
- "loss": 1.287,
+ "loss": 1.1861,
  "step": 40000
  },
- {
- "epoch": 4.406593406593407,
- "grad_norm": 6.07938814163208,
- "learning_rate": 2.3736263736263735e-05,
- "loss": 1.3094,
- "step": 40100
- },
- {
- "epoch": 4.417582417582418,
- "grad_norm": 5.3800530433654785,
- "learning_rate": 2.3296703296703297e-05,
- "loss": 1.2926,
- "step": 40200
- },
- {
- "epoch": 4.428571428571429,
- "grad_norm": 8.152463912963867,
- "learning_rate": 2.2857142857142858e-05,
- "loss": 1.2626,
- "step": 40300
- },
- {
- "epoch": 4.43956043956044,
- "grad_norm": 6.699965000152588,
- "learning_rate": 2.241758241758242e-05,
- "loss": 1.2763,
- "step": 40400
- },
  {
  "epoch": 4.450549450549451,
- "grad_norm": 6.094221591949463,
+ "grad_norm": 4.528479099273682,
  "learning_rate": 2.1978021978021977e-05,
- "loss": 1.2739,
+ "loss": 1.1999,
  "step": 40500
  },
- {
- "epoch": 4.461538461538462,
- "grad_norm": 5.403158187866211,
- "learning_rate": 2.1538461538461542e-05,
- "loss": 1.2842,
- "step": 40600
- },
- {
- "epoch": 4.472527472527473,
- "grad_norm": 6.363950252532959,
- "learning_rate": 2.10989010989011e-05,
- "loss": 1.2822,
- "step": 40700
- },
- {
- "epoch": 4.483516483516484,
- "grad_norm": 8.08597183227539,
- "learning_rate": 2.0659340659340658e-05,
- "loss": 1.2914,
- "step": 40800
- },
- {
- "epoch": 4.4945054945054945,
- "grad_norm": 5.750666618347168,
- "learning_rate": 2.0219780219780223e-05,
- "loss": 1.2919,
- "step": 40900
- },
  {
  "epoch": 4.5054945054945055,
- "grad_norm": 7.04209566116333,
+ "grad_norm": 4.554440021514893,
  "learning_rate": 1.978021978021978e-05,
- "loss": 1.2785,
+ "loss": 1.2045,
  "step": 41000
  },
- {
- "epoch": 4.516483516483516,
- "grad_norm": 4.417567253112793,
- "learning_rate": 1.9340659340659342e-05,
- "loss": 1.2891,
- "step": 41100
- },
- {
- "epoch": 4.527472527472527,
- "grad_norm": 8.629948616027832,
- "learning_rate": 1.89010989010989e-05,
- "loss": 1.3173,
- "step": 41200
- },
- {
- "epoch": 4.538461538461538,
- "grad_norm": 8.524527549743652,
- "learning_rate": 1.8461538461538465e-05,
- "loss": 1.2716,
- "step": 41300
- },
- {
- "epoch": 4.549450549450549,
- "grad_norm": 7.150579929351807,
- "learning_rate": 1.8021978021978023e-05,
- "loss": 1.2893,
- "step": 41400
- },
  {
  "epoch": 4.56043956043956,
- "grad_norm": 4.87885856628418,
+ "grad_norm": 4.417350769042969,
  "learning_rate": 1.7582417582417584e-05,
- "loss": 1.2788,
+ "loss": 1.2074,
  "step": 41500
  },
- {
- "epoch": 4.571428571428571,
- "grad_norm": 10.751927375793457,
- "learning_rate": 1.7142857142857145e-05,
- "loss": 1.2755,
- "step": 41600
- },
- {
- "epoch": 4.582417582417582,
- "grad_norm": 6.685251712799072,
- "learning_rate": 1.6703296703296703e-05,
- "loss": 1.2961,
- "step": 41700
- },
- {
- "epoch": 4.593406593406593,
- "grad_norm": 6.694922924041748,
- "learning_rate": 1.6263736263736265e-05,
- "loss": 1.2787,
- "step": 41800
- },
- {
- "epoch": 4.604395604395604,
- "grad_norm": 7.051819324493408,
- "learning_rate": 1.5824175824175826e-05,
- "loss": 1.2991,
- "step": 41900
- },
  {
  "epoch": 4.615384615384615,
- "grad_norm": 6.575582027435303,
+ "grad_norm": 4.048484802246094,
  "learning_rate": 1.5384615384615387e-05,
- "loss": 1.2925,
+ "loss": 1.203,
  "step": 42000
  },
- {
- "epoch": 4.626373626373626,
- "grad_norm": 4.073245525360107,
- "learning_rate": 1.4945054945054945e-05,
- "loss": 1.264,
- "step": 42100
- },
- {
- "epoch": 4.637362637362637,
- "grad_norm": 5.903039455413818,
- "learning_rate": 1.4505494505494508e-05,
- "loss": 1.2788,
- "step": 42200
- },
- {
- "epoch": 4.648351648351649,
- "grad_norm": 5.781927585601807,
- "learning_rate": 1.4065934065934066e-05,
- "loss": 1.2667,
- "step": 42300
- },
- {
- "epoch": 4.65934065934066,
- "grad_norm": 8.685062408447266,
- "learning_rate": 1.3626373626373626e-05,
- "loss": 1.2774,
- "step": 42400
- },
  {
  "epoch": 4.670329670329671,
- "grad_norm": 5.9609503746032715,
+ "grad_norm": 4.659012317657471,
  "learning_rate": 1.3186813186813187e-05,
- "loss": 1.2801,
+ "loss": 1.1917,
  "step": 42500
  },
- {
- "epoch": 4.681318681318682,
- "grad_norm": 5.177548885345459,
- "learning_rate": 1.2747252747252747e-05,
- "loss": 1.2885,
- "step": 42600
- },
- {
- "epoch": 4.6923076923076925,
- "grad_norm": 5.23708963394165,
- "learning_rate": 1.230769230769231e-05,
- "loss": 1.2728,
- "step": 42700
- },
- {
- "epoch": 4.7032967032967035,
- "grad_norm": 6.084412574768066,
- "learning_rate": 1.1868131868131868e-05,
- "loss": 1.2841,
- "step": 42800
- },
- {
- "epoch": 4.714285714285714,
- "grad_norm": 7.0549492835998535,
- "learning_rate": 1.1428571428571429e-05,
- "loss": 1.2945,
- "step": 42900
- },
  {
  "epoch": 4.725274725274725,
- "grad_norm": 6.189640998840332,
+ "grad_norm": 3.7058045864105225,
  "learning_rate": 1.0989010989010989e-05,
- "loss": 1.2607,
+ "loss": 1.1981,
  "step": 43000
  },
- {
- "epoch": 4.736263736263736,
- "grad_norm": 6.6250128746032715,
- "learning_rate": 1.054945054945055e-05,
- "loss": 1.2687,
- "step": 43100
- },
- {
- "epoch": 4.747252747252747,
- "grad_norm": 7.570830345153809,
- "learning_rate": 1.0109890109890111e-05,
- "loss": 1.2593,
- "step": 43200
- },
- {
- "epoch": 4.758241758241758,
- "grad_norm": 7.158848285675049,
- "learning_rate": 9.670329670329671e-06,
- "loss": 1.3019,
- "step": 43300
- },
- {
- "epoch": 4.769230769230769,
- "grad_norm": 4.113587856292725,
- "learning_rate": 9.230769230769232e-06,
- "loss": 1.2655,
- "step": 43400
- },
  {
  "epoch": 4.78021978021978,
- "grad_norm": 7.132826328277588,
+ "grad_norm": 3.214338779449463,
  "learning_rate": 8.791208791208792e-06,
- "loss": 1.2672,
+ "loss": 1.1868,
  "step": 43500
  },
- {
- "epoch": 4.791208791208791,
- "grad_norm": 6.412448406219482,
- "learning_rate": 8.351648351648352e-06,
- "loss": 1.2663,
- "step": 43600
- },
- {
- "epoch": 4.802197802197802,
- "grad_norm": 6.107234001159668,
- "learning_rate": 7.912087912087913e-06,
- "loss": 1.2722,
- "step": 43700
- },
- {
- "epoch": 4.813186813186813,
- "grad_norm": 10.841856002807617,
- "learning_rate": 7.4725274725274726e-06,
- "loss": 1.2749,
- "step": 43800
- },
- {
- "epoch": 4.824175824175824,
- "grad_norm": 5.174416542053223,
- "learning_rate": 7.032967032967033e-06,
- "loss": 1.2589,
- "step": 43900
- },
  {
  "epoch": 4.835164835164835,
- "grad_norm": 8.20840835571289,
+ "grad_norm": 3.2473630905151367,
  "learning_rate": 6.5934065934065935e-06,
- "loss": 1.3053,
+ "loss": 1.1979,
  "step": 44000
  },
- {
- "epoch": 4.846153846153846,
- "grad_norm": 6.337531089782715,
- "learning_rate": 6.153846153846155e-06,
- "loss": 1.2489,
- "step": 44100
- },
- {
- "epoch": 4.857142857142857,
- "grad_norm": 4.3771562576293945,
- "learning_rate": 5.7142857142857145e-06,
- "loss": 1.2597,
- "step": 44200
- },
- {
- "epoch": 4.868131868131869,
- "grad_norm": 4.407495021820068,
- "learning_rate": 5.274725274725275e-06,
- "loss": 1.2695,
- "step": 44300
- },
- {
- "epoch": 4.8791208791208796,
- "grad_norm": 6.802704811096191,
- "learning_rate": 4.8351648351648355e-06,
- "loss": 1.2716,
- "step": 44400
- },
  {
  "epoch": 4.8901098901098905,
- "grad_norm": 6.55114221572876,
+ "grad_norm": 3.6867451667785645,
  "learning_rate": 4.395604395604396e-06,
- "loss": 1.2531,
+ "loss": 1.1873,
  "step": 44500
  },
- {
- "epoch": 4.9010989010989015,
- "grad_norm": 6.069747447967529,
- "learning_rate": 3.9560439560439565e-06,
- "loss": 1.2504,
- "step": 44600
- },
- {
- "epoch": 4.912087912087912,
- "grad_norm": 6.654344081878662,
- "learning_rate": 3.5164835164835165e-06,
- "loss": 1.2859,
- "step": 44700
- },
- {
- "epoch": 4.923076923076923,
- "grad_norm": 5.364152431488037,
- "learning_rate": 3.0769230769230774e-06,
- "loss": 1.2886,
- "step": 44800
- },
- {
- "epoch": 4.934065934065934,
- "grad_norm": 6.0342607498168945,
- "learning_rate": 2.6373626373626375e-06,
- "loss": 1.2876,
- "step": 44900
- },
  {
  "epoch": 4.945054945054945,
- "grad_norm": 5.929255962371826,
+ "grad_norm": 4.085856914520264,
  "learning_rate": 2.197802197802198e-06,
- "loss": 1.3,
+ "loss": 1.1997,
  "step": 45000
  },
- {
- "epoch": 4.956043956043956,
- "grad_norm": 5.942348480224609,
- "learning_rate": 1.7582417582417583e-06,
- "loss": 1.2745,
- "step": 45100
- },
- {
- "epoch": 4.967032967032967,
- "grad_norm": 7.537661075592041,
- "learning_rate": 1.3186813186813187e-06,
- "loss": 1.2399,
- "step": 45200
- },
- {
- "epoch": 4.978021978021978,
- "grad_norm": 6.031464576721191,
- "learning_rate": 8.791208791208791e-07,
- "loss": 1.2424,
- "step": 45300
- },
- {
- "epoch": 4.989010989010989,
- "grad_norm": 6.415643692016602,
- "learning_rate": 4.3956043956043957e-07,
- "loss": 1.2585,
- "step": 45400
- },
  {
  "epoch": 5.0,
- "grad_norm": 6.854482173919678,
+ "grad_norm": 3.7809948921203613,
  "learning_rate": 0.0,
- "loss": 1.2887,
+ "loss": 1.1797,
  "step": 45500
  }
  ],
- "logging_steps": 100,
+ "logging_steps": 500,
  "max_steps": 45500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
@@ -3211,7 +663,7 @@
  "attributes": {}
  }
  },
- "total_flos": 3.250821618953257e+18,
+ "total_flos": 3.6063170257526784e+18,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:394886f47e018e532389708148cc7b102802e0b30e32db2484fb4b4f3ac88ae7
+oid sha256:09c8450b96bf0aa312a253c10156ec3e2c10a6f8f4c4de9d82477109290b625d
 size 5496