Training in progress, epoch 0, checkpoint

last-checkpoint/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:95d60326c2fb5bdc2227ea23053e3756d83dc36ddea29eca5dc260046692f87a
+oid sha256:b57425fe0979d64ee2759d723efa2d4501cd5b152cfaae9163875752c74cb054
 size 1172343536

last-checkpoint/global_step550/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f242f729d3b237c407cb89e280fb8792adf4462c6ee00a93454f157ca574eeb
+size 883824229

last-checkpoint/global_step550/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5629939c1c2380cc65c535d35dec5ee2eac687f266864fa91302a8d1e86add08
+size 883824293

last-checkpoint/global_step550/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ce6deea2b64c91d307a73284c004fd05960111e4fbb7676bd346623369064e6c
+size 883824293

last-checkpoint/global_step550/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6a26916dfaa62f83c848fae999d9044adf93863da585c54f7de2c45643d90c82
+size 883824293

last-checkpoint/global_step550/mp_rank_00_model_states.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d02b42c1375d7cc0a2cf3a976e266078049d353b0bdad6c66640869635ee225
+size 1172522073

last-checkpoint/latest

@@ -1 +1 @@
-global_step450
+global_step550

last-checkpoint/rng_state_0.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:14e0114f828a13cb0ca5e64776036e4737fc28bdf952551895db70f4b43b75fb
+oid sha256:ed8ed81af49ff602ccd13d5e84967bee6bca6a78ea6d7206a2104463b69a1059
 size 15429

last-checkpoint/rng_state_1.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6a1299b80d6a33e2be3e9ef939d892cd00826cf75bf71605c4233c7da79f587c
+oid sha256:f53e27b1f587841700b654201a4c9a9680a6d49a754e25cd2a82432e613117b4
 size 15429

last-checkpoint/rng_state_2.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:53a7808ab1ff80dd1fb0aa15bd8b839fe93e027e522f673e2233ef780746be2f
+oid sha256:db90f184c98dee2e965ee13d085db00953c6b696c1ef3da2b032f72d72e768a0
 size 15429

last-checkpoint/rng_state_3.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:829ff85ee41653334c647e740ab9be86f7a1e498563365c34f3b1d0c23c443af
+oid sha256:d32bc21799c1730e164d6b71e088e1ad7f9f709b9172577b8ff961df3d429692
 size 15429

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8693d7d5a684e7cf028814f0eb1563a3103edf4073a4bae9acefa6ff4db05daf
+oid sha256:6a96d6b66528c6387aa2a111a67d7f07cc9d6463a1b49e77f2099de0ba990f1c
 size 1401

last-checkpoint/trainer_state.json

@@ -1,10 +1,10 @@
 {
   "best_global_step": null,
-  "best_metric": 0.04127497971057892,
+  "best_metric": 0.03315730020403862,
   "best_model_checkpoint": null,
-  "epoch": 0.39087947882736157,
+  "epoch": 0.4777415852334419,
   "eval_steps": 50,
-  "global_step": 450,
+  "global_step": 550,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -1502,6 +1502,338 @@
       "eval_samples_per_second": 45.326,
       "eval_steps_per_second": 2.841,
       "step": 450
+    },
+    {
+      "epoch": 0.3952225841476656,
+      "grad_norm": 2.4726035594940186,
+      "learning_rate": 2.590891475251169e-05,
+      "logits/chosen": -0.265625,
+      "logits/rejected": -0.714550793170929,
+      "logps/chosen": -99.17500305175781,
+      "logps/rejected": -193.0500030517578,
+      "loss": 0.0333,
+      "rewards/accuracies": 0.9859374761581421,
+      "rewards/chosen": -2.3949217796325684,
+      "rewards/margins": 12.379687309265137,
+      "rewards/rejected": -14.782812118530273,
+      "step": 455
+    },
+    {
+      "epoch": 0.3995656894679696,
+      "grad_norm": 1.7090651988983154,
+      "learning_rate": 2.590658707847458e-05,
+      "logits/chosen": -0.23100586235523224,
+      "logits/rejected": -0.6619628667831421,
+      "logps/chosen": -91.0374984741211,
+      "logps/rejected": -180.9375,
+      "loss": 0.0458,
+      "rewards/accuracies": 0.9828125238418579,
+      "rewards/chosen": -0.9396575689315796,
+      "rewards/margins": 11.518750190734863,
+      "rewards/rejected": -12.459375381469727,
+      "step": 460
+    },
+    {
+      "epoch": 0.40390879478827363,
+      "grad_norm": 1.3948291540145874,
+      "learning_rate": 2.5904226900571074e-05,
+      "logits/chosen": -0.19654540717601776,
+      "logits/rejected": -0.6834961175918579,
+      "logps/chosen": -88.9375,
+      "logps/rejected": -183.75,
+      "loss": 0.035,
+      "rewards/accuracies": 0.9921875,
+      "rewards/chosen": -0.678784191608429,
+      "rewards/margins": 11.9453125,
+      "rewards/rejected": -12.620312690734863,
+      "step": 465
+    },
+    {
+      "epoch": 0.40825190010857765,
+      "grad_norm": 1.7862852811813354,
+      "learning_rate": 2.5901834224742923e-05,
+      "logits/chosen": -0.2498779296875,
+      "logits/rejected": -0.702197253704071,
+      "logps/chosen": -94.7125015258789,
+      "logps/rejected": -188.72500610351562,
+      "loss": 0.0374,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/chosen": -1.1595947742462158,
+      "rewards/margins": 11.94921875,
+      "rewards/rejected": -13.107812881469727,
+      "step": 470
+    },
+    {
+      "epoch": 0.41259500542888167,
+      "grad_norm": 2.126814365386963,
+      "learning_rate": 2.5899409057013675e-05,
+      "logits/chosen": -0.23180541396141052,
+      "logits/rejected": -0.739208996295929,
+      "logps/chosen": -93.05000305175781,
+      "logps/rejected": -191.22500610351562,
+      "loss": 0.0376,
+      "rewards/accuracies": 0.984375,
+      "rewards/chosen": -1.4010009765625,
+      "rewards/margins": 12.748437881469727,
+      "rewards/rejected": -14.1484375,
+      "step": 475
+    },
+    {
+      "epoch": 0.4169381107491857,
+      "grad_norm": 1.0905718803405762,
+      "learning_rate": 2.5896951403488697e-05,
+      "logits/chosen": -0.2298145294189453,
+      "logits/rejected": -0.7357422113418579,
+      "logps/chosen": -90.625,
+      "logps/rejected": -186.9250030517578,
+      "loss": 0.0479,
+      "rewards/accuracies": 0.981249988079071,
+      "rewards/chosen": -1.0264708995819092,
+      "rewards/margins": 12.8125,
+      "rewards/rejected": -13.842187881469727,
+      "step": 480
+    },
+    {
+      "epoch": 0.4212812160694897,
+      "grad_norm": 1.7384401559829712,
+      "learning_rate": 2.5894461270355127e-05,
+      "logits/chosen": -0.20979881286621094,
+      "logits/rejected": -0.714160144329071,
+      "logps/chosen": -89.80000305175781,
+      "logps/rejected": -180.89999389648438,
+      "loss": 0.047,
+      "rewards/accuracies": 0.979687511920929,
+      "rewards/chosen": -0.5426025390625,
+      "rewards/margins": 12.270312309265137,
+      "rewards/rejected": -12.817187309265137,
+      "step": 485
+    },
+    {
+      "epoch": 0.4256243213897937,
+      "grad_norm": 0.6238383650779724,
+      "learning_rate": 2.589193866388186e-05,
+      "logits/chosen": -0.19962844252586365,
+      "logits/rejected": -0.7691406011581421,
+      "logps/chosen": -86.78125,
+      "logps/rejected": -185.0749969482422,
+      "loss": 0.0246,
+      "rewards/accuracies": 0.9921875,
+      "rewards/chosen": -0.3228210508823395,
+      "rewards/margins": 12.751562118530273,
+      "rewards/rejected": -13.064062118530273,
+      "step": 490
+    },
+    {
+      "epoch": 0.42996742671009774,
+      "grad_norm": 0.9955658912658691,
+      "learning_rate": 2.5889383590419572e-05,
+      "logits/chosen": -0.28575438261032104,
+      "logits/rejected": -0.805957019329071,
+      "logps/chosen": -91.2125015258789,
+      "logps/rejected": -189.5749969482422,
+      "loss": 0.031,
+      "rewards/accuracies": 0.984375,
+      "rewards/chosen": -1.1411254405975342,
+      "rewards/margins": 12.951562881469727,
+      "rewards/rejected": -14.0859375,
+      "step": 495
+    },
+    {
+      "epoch": 0.43431053203040176,
+      "grad_norm": 2.1881649494171143,
+      "learning_rate": 2.588679605640065e-05,
+      "logits/chosen": -0.39592283964157104,
+      "logits/rejected": -0.893847644329071,
+      "logps/chosen": -101.30000305175781,
+      "logps/rejected": -200.5500030517578,
+      "loss": 0.0452,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/chosen": -2.7685546875,
+      "rewards/margins": 13.3125,
+      "rewards/rejected": -16.075000762939453,
+      "step": 500
+    },
+    {
+      "epoch": 0.43431053203040176,
+      "eval_logits/chosen": -0.4309884011745453,
+      "eval_logits/rejected": -0.9316617846488953,
+      "eval_logps/chosen": -103.70216369628906,
+      "eval_logps/rejected": -204.1552276611328,
+      "eval_loss": 0.04523569345474243,
+      "eval_rewards/accuracies": 0.9824007153511047,
+      "eval_rewards/chosen": -3.1199233531951904,
+      "eval_rewards/margins": 13.682310104370117,
+      "eval_rewards/rejected": -16.7991886138916,
+      "eval_runtime": 97.7314,
+      "eval_samples_per_second": 45.226,
+      "eval_steps_per_second": 2.834,
+      "step": 500
+    },
+    {
+      "epoch": 0.4386536373507057,
+      "grad_norm": 1.507201910018921,
+      "learning_rate": 2.588417606833921e-05,
+      "logits/chosen": -0.4728027284145355,
+      "logits/rejected": -0.9813476800918579,
+      "logps/chosen": -107.1875,
+      "logps/rejected": -211.97500610351562,
+      "loss": 0.0367,
+      "rewards/accuracies": 0.987500011920929,
+      "rewards/chosen": -3.632031202316284,
+      "rewards/margins": 14.167187690734863,
+      "rewards/rejected": -17.790624618530273,
+      "step": 505
+    },
+    {
+      "epoch": 0.44299674267100975,
+      "grad_norm": 0.9263981580734253,
+      "learning_rate": 2.588152363283106e-05,
+      "logits/chosen": -0.511279284954071,
+      "logits/rejected": -0.9991210699081421,
+      "logps/chosen": -114.6875,
+      "logps/rejected": -209.14999389648438,
+      "loss": 0.0682,
+      "rewards/accuracies": 0.979687511920929,
+      "rewards/chosen": -5.246874809265137,
+      "rewards/margins": 12.534375190734863,
+      "rewards/rejected": -17.779687881469727,
+      "step": 510
+    },
+    {
+      "epoch": 0.44733984799131377,
+      "grad_norm": 0.7554839253425598,
+      "learning_rate": 2.5878838756553715e-05,
+      "logits/chosen": -0.5076659917831421,
+      "logits/rejected": -0.9681640863418579,
+      "logps/chosen": -121.92500305175781,
+      "logps/rejected": -211.375,
+      "loss": 0.038,
+      "rewards/accuracies": 0.989062488079071,
+      "rewards/chosen": -6.292187690734863,
+      "rewards/margins": 11.65625,
+      "rewards/rejected": -17.9453125,
+      "step": 515
+    },
+    {
+      "epoch": 0.4516829533116178,
+      "grad_norm": 0.9180288910865784,
+      "learning_rate": 2.587612144626634e-05,
+      "logits/chosen": -0.4847168028354645,
+      "logits/rejected": -0.958984375,
+      "logps/chosen": -125.42500305175781,
+      "logps/rejected": -218.39999389648438,
+      "loss": 0.0349,
+      "rewards/accuracies": 0.984375,
+      "rewards/chosen": -7.428906440734863,
+      "rewards/margins": 11.963281631469727,
+      "rewards/rejected": -19.387500762939453,
+      "step": 520
+    },
+    {
+      "epoch": 0.4560260586319218,
+      "grad_norm": 0.5909751653671265,
+      "learning_rate": 2.5873371708809763e-05,
+      "logits/chosen": -0.5091308355331421,
+      "logits/rejected": -0.957324206829071,
+      "logps/chosen": -129.16250610351562,
+      "logps/rejected": -230.875,
+      "loss": 0.0351,
+      "rewards/accuracies": 0.9906250238418579,
+      "rewards/chosen": -8.124218940734863,
+      "rewards/margins": 13.762499809265137,
+      "rewards/rejected": -21.884374618530273,
+      "step": 525
+    },
+    {
+      "epoch": 0.4603691639522258,
+      "grad_norm": 2.2423148155212402,
+      "learning_rate": 2.5870589551106445e-05,
+      "logits/chosen": -0.6045898199081421,
+      "logits/rejected": -1.0859375,
+      "logps/chosen": -127.5,
+      "logps/rejected": -233.77499389648438,
+      "loss": 0.0534,
+      "rewards/accuracies": 0.981249988079071,
+      "rewards/chosen": -7.536718845367432,
+      "rewards/margins": 14.676562309265137,
+      "rewards/rejected": -22.212499618530273,
+      "step": 530
+    },
+    {
+      "epoch": 0.46471226927252984,
+      "grad_norm": 1.3651999235153198,
+      "learning_rate": 2.5867774980160467e-05,
+      "logits/chosen": -0.717089831829071,
+      "logits/rejected": -1.1892578601837158,
+      "logps/chosen": -124.4625015258789,
+      "logps/rejected": -228.8249969482422,
+      "loss": 0.0193,
+      "rewards/accuracies": 0.996874988079071,
+      "rewards/chosen": -6.423437595367432,
+      "rewards/margins": 14.548437118530273,
+      "rewards/rejected": -20.971874237060547,
+      "step": 535
+    },
+    {
+      "epoch": 0.46905537459283386,
+      "grad_norm": 1.8702569007873535,
+      "learning_rate": 2.5864928003057505e-05,
+      "logits/chosen": -0.762988269329071,
+      "logits/rejected": -1.267187476158142,
+      "logps/chosen": -113.23750305175781,
+      "logps/rejected": -224.39999389648438,
+      "loss": 0.0509,
+      "rewards/accuracies": 0.9781249761581421,
+      "rewards/chosen": -4.976171970367432,
+      "rewards/margins": 15.378125190734863,
+      "rewards/rejected": -20.356250762939453,
+      "step": 540
+    },
+    {
+      "epoch": 0.4733984799131379,
+      "grad_norm": 0.5270097851753235,
+      "learning_rate": 2.586204862696482e-05,
+      "logits/chosen": -0.71142578125,
+      "logits/rejected": -1.235937476158142,
+      "logps/chosen": -101.3499984741211,
+      "logps/rejected": -207.1750030517578,
+      "loss": 0.0294,
+      "rewards/accuracies": 0.9906250238418579,
+      "rewards/chosen": -3.095898389816284,
+      "rewards/margins": 14.653124809265137,
+      "rewards/rejected": -17.751562118530273,
+      "step": 545
+    },
+    {
+      "epoch": 0.4777415852334419,
+      "grad_norm": 0.6892856359481812,
+      "learning_rate": 2.5859136859131237e-05,
+      "logits/chosen": -0.674853503704071,
+      "logits/rejected": -1.1799805164337158,
+      "logps/chosen": -97.48750305175781,
+      "logps/rejected": -194.5500030517578,
+      "loss": 0.0303,
+      "rewards/accuracies": 0.9921875,
+      "rewards/chosen": -2.018017530441284,
+      "rewards/margins": 12.899999618530273,
+      "rewards/rejected": -14.918749809265137,
+      "step": 550
+    },
+    {
+      "epoch": 0.4777415852334419,
+      "eval_logits/chosen": -0.628990888595581,
+      "eval_logits/rejected": -1.142712116241455,
+      "eval_logps/chosen": -95.65342712402344,
+      "eval_logps/rejected": -188.6353759765625,
+      "eval_loss": 0.03315730020403862,
+      "eval_rewards/accuracies": 0.9875902533531189,
+      "eval_rewards/chosen": -1.6455535888671875,
+      "eval_rewards/margins": 12.3125,
+      "eval_rewards/rejected": -13.958709716796875,
+      "eval_runtime": 97.5044,
+      "eval_samples_per_second": 45.331,
+      "eval_steps_per_second": 2.841,
+      "step": 550
     }
   ],
   "logging_steps": 5,