Model save

README.md

@@ -2,7 +2,6 @@
 license: cc-by-nc-4.0
 base_model: facebook/nllb-200-3.3B
 tags:
-- alignment-handbook
 - generated_from_trainer
 model-index:
 - name: nllb-200-3.3B-full

all_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 2.9978431433840766,
-    "total_flos": 2.5671411434264986e+17,
-    "train_loss": 0.35637976608556393,
-    "train_runtime": 44055.9239,
+    "epoch": 2.9985620955893846,
+    "total_flos": 2.5672199949139968e+17,
+    "train_loss": 0.13155911801694273,
+    "train_runtime": 22482.9111,
     "train_samples": 63982,
-    "train_samples_per_second": 4.357,
-    "train_steps_per_second": 0.068
+    "train_samples_per_second": 8.537,
+    "train_steps_per_second": 0.133
 }

model-00001-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:854aad1436b17db9dda9a3bbfb34c8593c3fc5256160810ee11b76c9f8c378e9
+oid sha256:92e8002f9d7ad91add2a5a6fa943a1822f82dc5c00073840cbfcadd218b29ef5
 size 4986088344

model-00002-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bdac02ba51ca9accc1c112fbfc007ed5d7e1125d039a3349512e98570c5c19e0
+oid sha256:1b2f4fa931b7db9bdbb640c0539c07148676dc4a1b4ab8e985e59b4f00ab56ce
 size 4985688360

model-00003-of-00003.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:27aa0c57680f4ef6f513018cde1b956c93fe11d9ce8eddfd41c55847c5cfeb10
+oid sha256:f3bafae39e6d8883222c8ae788af5ebc939df144eb36034bff6ff0c8db53e3ad
 size 3407796744

train_results.json

@@ -1,9 +1,9 @@
 {
-    "epoch": 2.9978431433840766,
-    "total_flos": 2.5671411434264986e+17,
-    "train_loss": 0.35637976608556393,
-    "train_runtime": 44055.9239,
+    "epoch": 2.9985620955893846,
+    "total_flos": 2.5672199949139968e+17,
+    "train_loss": 0.13155911801694273,
+    "train_runtime": 22482.9111,
     "train_samples": 63982,
-    "train_samples_per_second": 4.357,
-    "train_steps_per_second": 0.068
+    "train_samples_per_second": 8.537,
+    "train_steps_per_second": 0.133
 }

trainer_state.json

@@ -1,7 +1,7 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 2.9978431433840766,
+  "epoch": 2.9985620955893846,
   "eval_steps": 500,
   "global_step": 2997,
   "is_hyper_param_search": false,
@@ -10,425 +10,425 @@
   "log_history": [
     {
       "epoch": 0.0500140664561908,
-      "grad_norm": 0.9210330247879028,
-      "learning_rate": 3.3333333333333333e-06,
-      "loss": 1.561,
+      "grad_norm": 1.5154740810394287,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.6338,
       "step": 50
     },
     {
       "epoch": 0.1000281329123816,
-      "grad_norm": 0.6426990032196045,
-      "learning_rate": 6.666666666666667e-06,
-      "loss": 0.9595,
+      "grad_norm": 0.7568167448043823,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.1959,
       "step": 100
     },
     {
       "epoch": 0.1500421993685724,
-      "grad_norm": 0.5717326998710632,
-      "learning_rate": 1e-05,
-      "loss": 0.6943,
+      "grad_norm": 0.6318637728691101,
+      "learning_rate": 5e-06,
+      "loss": 0.8316,
       "step": 150
     },
     {
       "epoch": 0.2000562658247632,
-      "grad_norm": 0.4604141414165497,
-      "learning_rate": 1.3333333333333333e-05,
-      "loss": 0.6171,
+      "grad_norm": 0.49124225974082947,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.7008,
       "step": 200
     },
     {
       "epoch": 0.25007033228095404,
-      "grad_norm": 0.45251232385635376,
-      "learning_rate": 1.6666666666666667e-05,
-      "loss": 0.5517,
+      "grad_norm": 0.49193131923675537,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.6234,
       "step": 250
     },
     {
       "epoch": 0.3000843987371448,
-      "grad_norm": 0.5194812417030334,
-      "learning_rate": 2e-05,
-      "loss": 0.523,
+      "grad_norm": 0.5137419700622559,
+      "learning_rate": 1e-05,
+      "loss": 0.5887,
       "step": 300
     },
     {
       "epoch": 0.35009846519333565,
-      "grad_norm": 0.4478510618209839,
-      "learning_rate": 1.9983043934122208e-05,
-      "loss": 0.4911,
+      "grad_norm": 0.4855007231235504,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.5523,
       "step": 350
     },
     {
       "epoch": 0.4001125316495264,
-      "grad_norm": 0.3697112500667572,
-      "learning_rate": 1.9932233238122834e-05,
-      "loss": 0.4631,
+      "grad_norm": 0.4011429250240326,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.5204,
       "step": 400
     },
     {
       "epoch": 0.45012659810571726,
-      "grad_norm": 0.4671821892261505,
-      "learning_rate": 1.984774022190361e-05,
-      "loss": 0.4416,
+      "grad_norm": 0.4894815683364868,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.4922,
       "step": 450
     },
     {
       "epoch": 0.5001406645619081,
-      "grad_norm": 0.3720390498638153,
-      "learning_rate": 1.972985141929439e-05,
-      "loss": 0.4219,
+      "grad_norm": 0.4187089204788208,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.4676,
       "step": 500
     },
     {
       "epoch": 0.5501547310180989,
-      "grad_norm": 0.3556453287601471,
-      "learning_rate": 1.9578966616355823e-05,
-      "loss": 0.4144,
+      "grad_norm": 0.37147971987724304,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.4564,
       "step": 550
     },
     {
       "epoch": 0.6001687974742896,
-      "grad_norm": 0.34724870324134827,
-      "learning_rate": 1.9395597495619634e-05,
-      "loss": 0.4129,
+      "grad_norm": 0.36486196517944336,
+      "learning_rate": 2e-05,
+      "loss": 0.451,
       "step": 600
     },
     {
       "epoch": 0.6501828639304804,
-      "grad_norm": 0.35355138778686523,
-      "learning_rate": 1.918036590086405e-05,
-      "loss": 0.3979,
+      "grad_norm": 0.39784976840019226,
+      "learning_rate": 1.999576008468646e-05,
+      "loss": 0.4313,
       "step": 650
     },
     {
       "epoch": 0.7001969303866713,
-      "grad_norm": 0.44766274094581604,
-      "learning_rate": 1.8934001728309003e-05,
-      "loss": 0.3843,
+      "grad_norm": 0.4364568293094635,
+      "learning_rate": 1.9983043934122208e-05,
+      "loss": 0.4132,
       "step": 700
     },
     {
       "epoch": 0.7502109968428621,
-      "grad_norm": 0.3572178781032562,
-      "learning_rate": 1.865734045138245e-05,
-      "loss": 0.3854,
+      "grad_norm": 0.7991167306900024,
+      "learning_rate": 1.9961862331387545e-05,
+      "loss": 0.4103,
       "step": 750
     },
     {
       "epoch": 0.8002250632990529,
-      "grad_norm": 0.37280118465423584,
-      "learning_rate": 1.8351320287451865e-05,
-      "loss": 0.3699,
+      "grad_norm": 0.3835698068141937,
+      "learning_rate": 1.9932233238122834e-05,
+      "loss": 0.3919,
       "step": 800
     },
     {
       "epoch": 0.8502391297552436,
-      "grad_norm": 0.3944127559661865,
-      "learning_rate": 1.8016979016129164e-05,
-      "loss": 0.3646,
+      "grad_norm": 0.40175750851631165,
+      "learning_rate": 1.9894181779297323e-05,
+      "loss": 0.3846,
       "step": 850
     },
     {
       "epoch": 0.9002531962114345,
-      "grad_norm": 0.32988861203193665,
-      "learning_rate": 1.7655450459938786e-05,
-      "loss": 0.3537,
+      "grad_norm": 0.3462737202644348,
+      "learning_rate": 1.984774022190361e-05,
+      "loss": 0.3712,
       "step": 900
     },
     {
       "epoch": 0.9502672626676253,
-      "grad_norm": 0.31536394357681274,
-      "learning_rate": 1.726796063928382e-05,
-      "loss": 0.3514,
+      "grad_norm": 0.31011104583740234,
+      "learning_rate": 1.9792947947595772e-05,
+      "loss": 0.3672,
       "step": 950
     },
     {
       "epoch": 1.0002813291238162,
-      "grad_norm": 0.3472963869571686,
-      "learning_rate": 1.6855823614749474e-05,
-      "loss": 0.3489,
+      "grad_norm": 0.34971198439598083,
+      "learning_rate": 1.972985141929439e-05,
+      "loss": 0.3635,
       "step": 1000
     },
     {
       "epoch": 1.050295395580007,
-      "grad_norm": 0.357003778219223,
-      "learning_rate": 1.6420437030843482e-05,
-      "loss": 0.3187,
+      "grad_norm": 0.3588427007198334,
+      "learning_rate": 1.9658504141786775e-05,
+      "loss": 0.3368,
       "step": 1050
     },
     {
       "epoch": 1.1003094620361977,
-      "grad_norm": 0.31990480422973633,
-      "learning_rate": 1.5963277376285646e-05,
-      "loss": 0.3117,
+      "grad_norm": 0.32608523964881897,
+      "learning_rate": 1.9578966616355823e-05,
+      "loss": 0.3279,
       "step": 1100
     },
     {
       "epoch": 1.1503235284923885,
-      "grad_norm": 0.3657567501068115,
-      "learning_rate": 1.5485894976919836e-05,
-      "loss": 0.3089,
+      "grad_norm": 0.3917510509490967,
+      "learning_rate": 1.9491306289475957e-05,
+      "loss": 0.3239,
       "step": 1150
     },
     {
       "epoch": 1.2003375949485793,
-      "grad_norm": 0.3026184141635895,
-      "learning_rate": 1.4989908738228567e-05,
-      "loss": 0.3005,
+      "grad_norm": 0.3702155649662018,
+      "learning_rate": 1.9395597495619634e-05,
+      "loss": 0.3147,
       "step": 1200
     },
     {
       "epoch": 1.25035166140477,
-      "grad_norm": 0.3388034999370575,
-      "learning_rate": 1.4477000655279376e-05,
-      "loss": 0.3042,
+      "grad_norm": 0.34320342540740967,
+      "learning_rate": 1.9291921394223e-05,
+      "loss": 0.3178,
       "step": 1250
     },
     {
       "epoch": 1.3003657278609608,
-      "grad_norm": 0.3384702205657959,
-      "learning_rate": 1.394891010872102e-05,
-      "loss": 0.3057,
+      "grad_norm": 0.38336554169654846,
+      "learning_rate": 1.918036590086405e-05,
+      "loss": 0.3195,
       "step": 1300
     },
     {
       "epoch": 1.3503797943171518,
-      "grad_norm": 0.3673734664916992,
-      "learning_rate": 1.3407427966172866e-05,
-      "loss": 0.2958,
+      "grad_norm": 0.4073985517024994,
+      "learning_rate": 1.90610256127117e-05,
+      "loss": 0.308,
       "step": 1350
     },
     {
       "epoch": 1.4003938607733426,
-      "grad_norm": 0.35079866647720337,
-      "learning_rate": 1.2854390509011061e-05,
-      "loss": 0.298,
+      "grad_norm": 0.37673142552375793,
+      "learning_rate": 1.8934001728309003e-05,
+      "loss": 0.3096,
       "step": 1400
     },
     {
       "epoch": 1.4504079272295334,
-      "grad_norm": 0.3334648013114929,
-      "learning_rate": 1.2291673205146908e-05,
-      "loss": 0.3008,
+      "grad_norm": 0.36145591735839844,
+      "learning_rate": 1.8799401961758492e-05,
+      "loss": 0.3121,
       "step": 1450
     },
     {
       "epoch": 1.5004219936857242,
-      "grad_norm": 0.3893981873989105,
-      "learning_rate": 1.1721184348915384e-05,
-      "loss": 0.2917,
+      "grad_norm": 0.4171595275402069,
+      "learning_rate": 1.865734045138245e-05,
+      "loss": 0.3017,
       "step": 1500
     },
     {
-      "epoch": 1.550436060141915,
-      "grad_norm": 0.31972965598106384,
+      "epoch": 1.5511550123472226,
+      "grad_norm": 0.32917237281799316,
       "learning_rate": 1.1144858589642251e-05,
-      "loss": 0.2926,
+      "loss": 0.2992,
       "step": 1550
     },
     {
-      "epoch": 1.6004501265981057,
-      "grad_norm": 0.31529781222343445,
+      "epoch": 1.6011690788034134,
+      "grad_norm": 0.3282073736190796,
       "learning_rate": 1.0564650370835772e-05,
-      "loss": 0.2876,
+      "loss": 0.293,
       "step": 1600
     },
     {
-      "epoch": 1.6504641930542965,
-      "grad_norm": 0.3786483705043793,
+      "epoch": 1.6511831452596044,
+      "grad_norm": 0.31431856751441956,
       "learning_rate": 9.982527302252135e-06,
-      "loss": 0.2904,
+      "loss": 0.2969,
       "step": 1650
     },
     {
-      "epoch": 1.7004782595104873,
-      "grad_norm": 0.33780983090400696,
+      "epoch": 1.7011972117157952,
+      "grad_norm": 0.3012774586677551,
       "learning_rate": 9.40046348731131e-06,
-      "loss": 0.2906,
+      "loss": 0.2954,
       "step": 1700
     },
     {
-      "epoch": 1.750492325966678,
-      "grad_norm": 0.3387848734855652,
+      "epoch": 1.751211278171986,
+      "grad_norm": 0.2984926104545593,
       "learning_rate": 8.820432828491542e-06,
-      "loss": 0.2833,
+      "loss": 0.2885,
       "step": 1750
     },
     {
-      "epoch": 1.8005063924228688,
-      "grad_norm": 0.280521422624588,
+      "epoch": 1.8012253446281767,
+      "grad_norm": 0.30490660667419434,
       "learning_rate": 8.244402333405252e-06,
-      "loss": 0.2779,
+      "loss": 0.2841,
       "step": 1800
     },
     {
-      "epoch": 1.8505204588790596,
-      "grad_norm": 0.3127559721469879,
+      "epoch": 1.8512394110843675,
+      "grad_norm": 0.3301903009414673,
       "learning_rate": 7.674325444256899e-06,
-      "loss": 0.2898,
+      "loss": 0.2935,
       "step": 1850
     },
     {
-      "epoch": 1.9005345253352506,
-      "grad_norm": 0.30676454305648804,
+      "epoch": 1.9012534775405583,
+      "grad_norm": 0.3045901954174042,
       "learning_rate": 7.112135413304042e-06,
-      "loss": 0.2813,
+      "loss": 0.2855,
       "step": 1900
     },
     {
-      "epoch": 1.9505485917914414,
-      "grad_norm": 0.31918200850486755,
+      "epoch": 1.951267543996749,
+      "grad_norm": 0.310871958732605,
       "learning_rate": 6.55973874678682e-06,
-      "loss": 0.2777,
+      "loss": 0.2831,
       "step": 1950
     },
     {
-      "epoch": 2.0005626582476324,
-      "grad_norm": 0.32422205805778503,
+      "epoch": 2.00128161045294,
+      "grad_norm": 0.3118513822555542,
       "learning_rate": 6.0190087395588596e-06,
-      "loss": 0.2767,
+      "loss": 0.2823,
       "step": 2000
     },
     {
-      "epoch": 2.050576724703823,
-      "grad_norm": 0.31339433789253235,
+      "epoch": 2.0512956769091306,
+      "grad_norm": 0.2910955250263214,
       "learning_rate": 5.491779122345093e-06,
-      "loss": 0.2571,
+      "loss": 0.2576,
       "step": 2050
     },
     {
-      "epoch": 2.100590791160014,
-      "grad_norm": 0.342579185962677,
+      "epoch": 2.1013097433653214,
+      "grad_norm": 0.3105012774467468,
       "learning_rate": 4.979837843169959e-06,
-      "loss": 0.2516,
+      "loss": 0.2522,
       "step": 2100
     },
     {
-      "epoch": 2.1506048576162047,
-      "grad_norm": 0.38084590435028076,
+      "epoch": 2.151323809821512,
+      "grad_norm": 0.31710630655288696,
       "learning_rate": 4.484921004044509e-06,
-      "loss": 0.2536,
+      "loss": 0.2546,
       "step": 2150
     },
     {
-      "epoch": 2.2006189240723955,
-      "grad_norm": 0.3659977316856384,
+      "epoch": 2.201337876277703,
+      "grad_norm": 0.41157081723213196,
       "learning_rate": 4.008706973474391e-06,
-      "loss": 0.2508,
+      "loss": 0.2531,
       "step": 2200
     },
     {
-      "epoch": 2.2506329905285862,
-      "grad_norm": 0.3062564730644226,
+      "epoch": 2.2513519427338937,
+      "grad_norm": 0.3166573941707611,
       "learning_rate": 3.5528106947544626e-06,
-      "loss": 0.2501,
+      "loss": 0.2519,
       "step": 2250
     },
     {
-      "epoch": 2.300647056984777,
-      "grad_norm": 0.31410741806030273,
+      "epoch": 2.301366009190085,
+      "grad_norm": 0.3242489695549011,
       "learning_rate": 3.118778209351808e-06,
-      "loss": 0.2555,
+      "loss": 0.254,
       "step": 2300
     },
     {
-      "epoch": 2.350661123440968,
-      "grad_norm": 0.3308572769165039,
+      "epoch": 2.3513800756462757,
+      "grad_norm": 0.34881240129470825,
       "learning_rate": 2.7080814139495402e-06,
-      "loss": 0.2519,
+      "loss": 0.2526,
       "step": 2350
     },
     {
-      "epoch": 2.4006751898971586,
-      "grad_norm": 0.32767045497894287,
+      "epoch": 2.4013941421024665,
+      "grad_norm": 0.3267311751842499,
       "learning_rate": 2.322113068931391e-06,
-      "loss": 0.2496,
+      "loss": 0.251,
       "step": 2400
     },
     {
-      "epoch": 2.4506892563533493,
-      "grad_norm": 0.302573025226593,
+      "epoch": 2.4514082085586573,
+      "grad_norm": 0.3213905394077301,
       "learning_rate": 1.9621820752343324e-06,
-      "loss": 0.245,
+      "loss": 0.246,
       "step": 2450
     },
     {
-      "epoch": 2.50070332280954,
-      "grad_norm": 0.29837408661842346,
+      "epoch": 2.501422275014848,
+      "grad_norm": 0.48464563488960266,
       "learning_rate": 1.629509035586484e-06,
-      "loss": 0.2532,
+      "loss": 0.2527,
       "step": 2500
     },
     {
-      "epoch": 2.550717389265731,
-      "grad_norm": 0.3348439931869507,
+      "epoch": 2.551436341471039,
+      "grad_norm": 0.31706663966178894,
       "learning_rate": 1.3252221151830513e-06,
-      "loss": 0.2457,
+      "loss": 0.2465,
       "step": 2550
     },
     {
-      "epoch": 2.6007314557219217,
-      "grad_norm": 0.37546366453170776,
+      "epoch": 2.6014504079272296,
+      "grad_norm": 0.33972597122192383,
       "learning_rate": 1.0503532158376584e-06,
-      "loss": 0.2475,
+      "loss": 0.2483,
       "step": 2600
     },
     {
-      "epoch": 2.6507455221781124,
-      "grad_norm": 0.2989293038845062,
+      "epoch": 2.6514644743834204,
+      "grad_norm": 0.2904072105884552,
       "learning_rate": 8.058344765833171e-07,
-      "loss": 0.246,
+      "loss": 0.2465,
       "step": 2650
     },
     {
-      "epoch": 2.7007595886343037,
-      "grad_norm": 0.32395079731941223,
+      "epoch": 2.701478540839611,
+      "grad_norm": 0.32852381467819214,
       "learning_rate": 5.924951125902545e-07,
-      "loss": 0.2495,
+      "loss": 0.2512,
       "step": 2700
     },
     {
-      "epoch": 2.7507736550904944,
-      "grad_norm": 0.31438368558883667,
+      "epoch": 2.751492607295802,
+      "grad_norm": 0.2948131263256073,
       "learning_rate": 4.11058603120511e-07,
-      "loss": 0.2476,
+      "loss": 0.2483,
       "step": 2750
     },
     {
-      "epoch": 2.800787721546685,
-      "grad_norm": 0.30735209584236145,
+      "epoch": 2.8015066737519927,
+      "grad_norm": 0.33766064047813416,
       "learning_rate": 2.6214023805552826e-07,
-      "loss": 0.2481,
+      "loss": 0.2496,
       "step": 2800
     },
     {
-      "epoch": 2.850801788002876,
-      "grad_norm": 0.3182600736618042,
+      "epoch": 2.8515207402081835,
+      "grad_norm": 0.29850533604621887,
       "learning_rate": 1.462450313169983e-07,
-      "loss": 0.245,
+      "loss": 0.2444,
       "step": 2850
     },
     {
-      "epoch": 2.9008158544590668,
-      "grad_norm": 0.3001386523246765,
+      "epoch": 2.9015348066643742,
+      "grad_norm": 0.33009734749794006,
       "learning_rate": 6.376600825699463e-08,
-      "loss": 0.2459,
+      "loss": 0.2478,
       "step": 2900
     },
     {
-      "epoch": 2.9508299209152575,
-      "grad_norm": 0.3104873299598694,
+      "epoch": 2.951548873120565,
+      "grad_norm": 0.5311033725738525,
       "learning_rate": 1.49828728252277e-08,
-      "loss": 0.2419,
+      "loss": 0.2412,
       "step": 2950
     },
     {
-      "epoch": 2.9978431433840766,
+      "epoch": 2.9985620955893846,
       "step": 2997,
-      "total_flos": 2.5671411434264986e+17,
-      "train_loss": 0.35637976608556393,
-      "train_runtime": 44055.9239,
-      "train_samples_per_second": 4.357,
-      "train_steps_per_second": 0.068
+      "total_flos": 2.5672199949139968e+17,
+      "train_loss": 0.13155911801694273,
+      "train_runtime": 22482.9111,
+      "train_samples_per_second": 8.537,
+      "train_steps_per_second": 0.133
     }
   ],
   "logging_steps": 50,
@@ -436,7 +436,7 @@
   "num_input_tokens_seen": 0,
   "num_train_epochs": 3,
   "save_steps": 100,
-  "total_flos": 2.5671411434264986e+17,
+  "total_flos": 2.5672199949139968e+17,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null