Training in progress, step 10, checkpoint

last-checkpoint/adapter_config.json

@@ -20,13 +20,13 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "k_proj",
+    "q_proj",
     "gate_proj",
-    "o_proj",
+    "k_proj",
     "down_proj",
     "v_proj",
     "up_proj",
-    "q_proj"
+    "o_proj"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

last-checkpoint/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:21cb1efe47a7b9ec7582f9c4cad57e9fdc96cddc43b5fe446da11199b86121df
+oid sha256:08801e765bddd96df04227d8b83cb7920b56ab21e0909367149debeab7cb2ebb
 size 83115256

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ab11797a5a48c8b5091a20e4ab8445d69bbcf0b52c3045370731dea57405fdcc
+oid sha256:47a18d76d4250392cb24be1e6c582f52939cd7de8bced38803dba85f973db406
 size 166351098

last-checkpoint/trainer_state.json

@@ -10,152 +10,152 @@
   "log_history": [
     {
       "epoch": 0.0005925925925925926,
-      "grad_norm": 10.896183013916016,
+      "grad_norm": 0.20394721627235413,
       "learning_rate": 5e-07,
       "logits/chosen": -4.444676399230957,
       "logits/rejected": -4.0909342765808105,
-      "logps/chosen": -186.1875,
-      "logps/rejected": -228.68560791015625,
-      "loss": 0.6931,
-      "rewards/accuracies": 0.0,
-      "rewards/chosen": 0.0,
-      "rewards/margins": 0.0,
-      "rewards/rejected": 0.0,
+      "logps/chosen": -3.4668707847595215,
+      "logps/rejected": -3.5732126235961914,
+      "loss": 0.7019,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": -0.34668707847595215,
+      "rewards/margins": 0.010634180158376694,
+      "rewards/rejected": -0.35732126235961914,
       "step": 1
     },
     {
       "epoch": 0.0011851851851851852,
-      "grad_norm": 7.562353610992432,
+      "grad_norm": 0.11779391765594482,
       "learning_rate": 4.849231551964771e-07,
       "logits/chosen": -4.530362606048584,
       "logits/rejected": -3.981240749359131,
-      "logps/chosen": -192.2100830078125,
-      "logps/rejected": -213.73086547851562,
-      "loss": 0.6931,
-      "rewards/accuracies": 0.0,
-      "rewards/chosen": 0.0,
-      "rewards/margins": 0.0,
-      "rewards/rejected": 0.0,
+      "logps/chosen": -3.0032825469970703,
+      "logps/rejected": -3.3395447731018066,
+      "loss": 0.6892,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": -0.30032825469970703,
+      "rewards/margins": 0.03362620621919632,
+      "rewards/rejected": -0.33395445346832275,
       "step": 2
     },
     {
       "epoch": 0.0017777777777777779,
-      "grad_norm": 13.934762001037598,
+      "grad_norm": 0.30934685468673706,
       "learning_rate": 4.415111107797445e-07,
-      "logits/chosen": -4.541503429412842,
-      "logits/rejected": -4.324827194213867,
-      "logps/chosen": -200.9736328125,
-      "logps/rejected": -224.6981658935547,
-      "loss": 0.6691,
-      "rewards/accuracies": 1.0,
-      "rewards/chosen": 0.022507095709443092,
-      "rewards/margins": 0.04895630106329918,
-      "rewards/rejected": -0.026449203491210938,
+      "logits/chosen": -4.551039695739746,
+      "logits/rejected": -4.332643508911133,
+      "logps/chosen": -3.143617868423462,
+      "logps/rejected": -3.5061092376708984,
+      "loss": 0.6877,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": -0.3143618106842041,
+      "rewards/margins": 0.03624912351369858,
+      "rewards/rejected": -0.3506109118461609,
       "step": 3
     },
     {
       "epoch": 0.0023703703703703703,
-      "grad_norm": 13.249699592590332,
+      "grad_norm": 0.38463467359542847,
       "learning_rate": 3.75e-07,
-      "logits/chosen": -3.9642434120178223,
-      "logits/rejected": -4.461881160736084,
-      "logps/chosen": -239.5960693359375,
-      "logps/rejected": -202.10887145996094,
-      "loss": 0.7259,
-      "rewards/accuracies": 0.0,
-      "rewards/chosen": -0.001570509746670723,
-      "rewards/margins": -0.06432266533374786,
-      "rewards/rejected": 0.06275215744972229,
+      "logits/chosen": -3.961301326751709,
+      "logits/rejected": -4.457886219024658,
+      "logps/chosen": -3.7372827529907227,
+      "logps/rejected": -6.128005027770996,
+      "loss": 0.6153,
+      "rewards/accuracies": 0.5,
+      "rewards/chosen": -0.37372830510139465,
+      "rewards/margins": 0.23907223343849182,
+      "rewards/rejected": -0.6128004789352417,
       "step": 4
     },
     {
       "epoch": 0.002962962962962963,
-      "grad_norm": 6.773025989532471,
+      "grad_norm": 0.28858768939971924,
       "learning_rate": 2.934120444167326e-07,
-      "logits/chosen": -4.136728286743164,
-      "logits/rejected": -4.700671672821045,
-      "logps/chosen": -169.1282958984375,
-      "logps/rejected": -149.28570556640625,
-      "loss": 0.6866,
-      "rewards/accuracies": 0.75,
-      "rewards/chosen": 0.017114639282226562,
-      "rewards/margins": 0.013147544115781784,
-      "rewards/rejected": 0.003967094700783491,
+      "logits/chosen": -4.138806343078613,
+      "logits/rejected": -4.704649925231934,
+      "logps/chosen": -2.6420888900756836,
+      "logps/rejected": -4.444168567657471,
+      "loss": 0.6272,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -0.2642088830471039,
+      "rewards/margins": 0.18020795285701752,
+      "rewards/rejected": -0.4444168508052826,
       "step": 5
     },
     {
       "epoch": 0.0035555555555555557,
-      "grad_norm": 13.418476104736328,
+      "grad_norm": 0.18824324011802673,
       "learning_rate": 2.065879555832674e-07,
-      "logits/chosen": -5.0562872886657715,
-      "logits/rejected": -5.145793914794922,
-      "logps/chosen": -194.10568237304688,
-      "logps/rejected": -202.68751525878906,
-      "loss": 0.6884,
-      "rewards/accuracies": 0.75,
-      "rewards/chosen": -0.034047700464725494,
-      "rewards/margins": 0.01024474948644638,
-      "rewards/rejected": -0.044292449951171875,
+      "logits/chosen": -5.056676864624023,
+      "logits/rejected": -5.151688098907471,
+      "logps/chosen": -3.0351624488830566,
+      "logps/rejected": -3.163884162902832,
+      "loss": 0.6995,
+      "rewards/accuracies": 0.5,
+      "rewards/chosen": -0.3035162687301636,
+      "rewards/margins": 0.012872166931629181,
+      "rewards/rejected": -0.31638842821121216,
       "step": 6
     },
     {
       "epoch": 0.004148148148148148,
-      "grad_norm": 6.973136901855469,
+      "grad_norm": 0.10379564017057419,
       "learning_rate": 1.2500000000000005e-07,
-      "logits/chosen": -4.244935035705566,
-      "logits/rejected": -3.681328058242798,
-      "logps/chosen": -176.23780822753906,
-      "logps/rejected": -222.976318359375,
-      "loss": 0.7073,
-      "rewards/accuracies": 0.5,
-      "rewards/chosen": 0.016452789306640625,
-      "rewards/margins": -0.027773665264248848,
-      "rewards/rejected": 0.044226452708244324,
+      "logits/chosen": -4.240752696990967,
+      "logits/rejected": -3.6786282062530518,
+      "logps/chosen": -2.834850788116455,
+      "logps/rejected": -3.4888226985931396,
+      "loss": 0.6739,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": -0.2834850549697876,
+      "rewards/margins": 0.0653972178697586,
+      "rewards/rejected": -0.3488822877407074,
       "step": 7
     },
     {
       "epoch": 0.004740740740740741,
-      "grad_norm": 10.38602066040039,
+      "grad_norm": 0.1659402847290039,
       "learning_rate": 5.848888922025552e-08,
-      "logits/chosen": -4.409693241119385,
-      "logits/rejected": -4.41237735748291,
-      "logps/chosen": -181.64944458007812,
-      "logps/rejected": -196.00564575195312,
-      "loss": 0.6896,
-      "rewards/accuracies": 0.5,
-      "rewards/chosen": 0.005864143371582031,
-      "rewards/margins": 0.007416536100208759,
-      "rewards/rejected": -0.001552392728626728,
+      "logits/chosen": -4.413921356201172,
+      "logits/rejected": -4.417206764221191,
+      "logps/chosen": -2.845391273498535,
+      "logps/rejected": -3.0666072368621826,
+      "loss": 0.6947,
+      "rewards/accuracies": 0.75,
+      "rewards/chosen": -0.284539133310318,
+      "rewards/margins": 0.022121611982584,
+      "rewards/rejected": -0.3066607415676117,
       "step": 8
     },
     {
       "epoch": 0.005333333333333333,
-      "grad_norm": 9.169404983520508,
+      "grad_norm": 0.8284716010093689,
       "learning_rate": 1.507684480352292e-08,
-      "logits/chosen": -4.903069972991943,
-      "logits/rejected": -4.5977277755737305,
-      "logps/chosen": -154.88751220703125,
-      "logps/rejected": -179.42715454101562,
-      "loss": 0.695,
-      "rewards/accuracies": 0.5,
-      "rewards/chosen": -0.016914749518036842,
-      "rewards/margins": -0.003633500775322318,
-      "rewards/rejected": -0.013281249441206455,
+      "logits/chosen": -4.9146857261657715,
+      "logits/rejected": -4.60509729385376,
+      "logps/chosen": -5.166627883911133,
+      "logps/rejected": -5.358448028564453,
+      "loss": 0.7251,
+      "rewards/accuracies": 0.25,
+      "rewards/chosen": -0.5166628360748291,
+      "rewards/margins": 0.019182000309228897,
+      "rewards/rejected": -0.5358448028564453,
       "step": 9
     },
     {
       "epoch": 0.005925925925925926,
-      "grad_norm": 10.337571144104004,
+      "grad_norm": 0.24434176087379456,
       "learning_rate": 0.0,
-      "logits/chosen": -3.817504644393921,
-      "logits/rejected": -3.697187900543213,
-      "logps/chosen": -233.053466796875,
-      "logps/rejected": -254.43130493164062,
-      "loss": 0.7085,
-      "rewards/accuracies": 0.5,
-      "rewards/chosen": -0.06154556944966316,
-      "rewards/margins": -0.029002761468291283,
-      "rewards/rejected": -0.03254280239343643,
+      "logits/chosen": -3.8119335174560547,
+      "logits/rejected": -3.695971727371216,
+      "logps/chosen": -3.6296451091766357,
+      "logps/rejected": -3.969270706176758,
+      "loss": 0.6888,
+      "rewards/accuracies": 1.0,
+      "rewards/chosen": -0.36296454071998596,
+      "rewards/margins": 0.03396253287792206,
+      "rewards/rejected": -0.3969270884990692,
       "step": 10
     }
   ],

last-checkpoint/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:265f72a5d7c9bd4d95478bba01ec7511ef89dd45479699ac483e45fdedd76cef
-size 5944
+oid sha256:b53a6ce3370b2ce423e2397cd1b929adc936cfb1905e6d876f1d8058230dac19
+size 5624