Upload folder using huggingface_hub

README.md

@@ -201,4 +201,4 @@ Carbon emissions can be estimated using the [Machine Learning Impact calculator]
 
 ### Framework versions
 
-- PEFT 0.9.1.dev0
+- PEFT 0.10.1.dev0

adapter_config.json

@@ -21,13 +21,13 @@
   "revision": null,
   "target_modules": [
     "q_proj",
-    "down_proj",
     "gate_proj",
-    "k_proj",
-    "lm_head",
+    "down_proj",
     "o_proj",
+    "up_proj",
     "v_proj",
-    "up_proj"
+    "k_proj",
+    "lm_head"
   ],
   "task_type": "CAUSAL_LM",
   "use_dora": false,

adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:881950e6663806a7a1cae84e16806fa4b0c356c7700d75b09e824d5a1038d104
+oid sha256:bb0de04c5fdc799c3eb184bbca9e949ace46bce96733c317f5c6109b8809a446
 size 1135046584

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5ee43ea073ea8bfd0122999e3f5401effb661df83831472b7c7613173f45d8e4
+oid sha256:cab7f8f12963691c3b150a8f1565ada324e212b008eb8dfbb4eb587989473aea
 size 172002140

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e1fab63bc4c064521d684a84ba72e6a51341affd3d4ae1cf57dc3dee1786bdeb
+oid sha256:728452d853406dd663b3b995f7585156461f2a67ebd775cfe7bf168c7e9352ff
 size 14244

scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1fca2dbc47f45142fd4d9074083bcceb2bd1672185349ccd98ffcef54047507e
+oid sha256:6328c51b1aa7a22a52317ddd7a925ee30a4741f0649ee5c9a4a201c53a1c6520
 size 1064

trainer_state.json

@@ -1,1557 +1,1497 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.9941125373998648,
+  "epoch": 0.9555062252678312,
   "eval_steps": 100,
-  "global_step": 10300,
+  "global_step": 9900,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.01,
-      "grad_norm": 5.765194416046143,
+      "grad_norm": 5.168999671936035,
       "learning_rate": 2.483498899926662e-05,
-      "loss": 1.8358,
+      "loss": 1.8368,
       "step": 100
     },
     {
       "epoch": 0.01,
-      "eval_loss": 1.4509096145629883,
-      "eval_runtime": 27.2685,
-      "eval_samples_per_second": 18.336,
-      "eval_steps_per_second": 2.31,
+      "eval_loss": 1.4510709047317505,
+      "eval_runtime": 27.2844,
+      "eval_samples_per_second": 18.325,
+      "eval_steps_per_second": 2.309,
       "step": 100
     },
     {
       "epoch": 0.02,
-      "grad_norm": 5.369307041168213,
+      "grad_norm": 5.417923927307129,
       "learning_rate": 2.466831122074805e-05,
-      "loss": 1.6055,
+      "loss": 1.6043,
       "step": 200
     },
     {
       "epoch": 0.02,
-      "eval_loss": 1.4317318201065063,
-      "eval_runtime": 27.2583,
-      "eval_samples_per_second": 18.343,
-      "eval_steps_per_second": 2.311,
+      "eval_loss": 1.4302723407745361,
+      "eval_runtime": 27.2442,
+      "eval_samples_per_second": 18.353,
+      "eval_steps_per_second": 2.312,
       "step": 200
     },
     {
       "epoch": 0.03,
-      "grad_norm": 5.935494422912598,
+      "grad_norm": 5.932116985321045,
       "learning_rate": 2.450163344222948e-05,
-      "loss": 1.7293,
+      "loss": 1.7302,
       "step": 300
     },
     {
       "epoch": 0.03,
-      "eval_loss": 1.4047609567642212,
-      "eval_runtime": 27.2345,
-      "eval_samples_per_second": 18.359,
+      "eval_loss": 1.4036405086517334,
+      "eval_runtime": 27.2418,
+      "eval_samples_per_second": 18.354,
       "eval_steps_per_second": 2.313,
       "step": 300
     },
     {
       "epoch": 0.04,
-      "grad_norm": 3.3592865467071533,
+      "grad_norm": 3.3278746604919434,
       "learning_rate": 2.4334955663710914e-05,
-      "loss": 1.59,
+      "loss": 1.5892,
       "step": 400
     },
     {
       "epoch": 0.04,
-      "eval_loss": 1.391677975654602,
-      "eval_runtime": 27.2416,
-      "eval_samples_per_second": 18.354,
+      "eval_loss": 1.3917973041534424,
+      "eval_runtime": 27.2382,
+      "eval_samples_per_second": 18.357,
       "eval_steps_per_second": 2.313,
       "step": 400
     },
     {
       "epoch": 0.05,
-      "grad_norm": 5.640110492706299,
+      "grad_norm": 5.543770790100098,
       "learning_rate": 2.4168277885192348e-05,
-      "loss": 1.6514,
+      "loss": 1.6516,
       "step": 500
     },
     {
       "epoch": 0.05,
-      "eval_loss": 1.3870183229446411,
-      "eval_runtime": 27.2222,
-      "eval_samples_per_second": 18.367,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.3847407102584839,
+      "eval_runtime": 27.2419,
+      "eval_samples_per_second": 18.354,
+      "eval_steps_per_second": 2.313,
       "step": 500
     },
     {
       "epoch": 0.06,
-      "grad_norm": 4.0497846603393555,
+      "grad_norm": 4.016234874725342,
       "learning_rate": 2.400160010667378e-05,
-      "loss": 1.5763,
+      "loss": 1.5757,
       "step": 600
     },
     {
       "epoch": 0.06,
-      "eval_loss": 1.3725069761276245,
-      "eval_runtime": 27.2203,
-      "eval_samples_per_second": 18.369,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.3731025457382202,
+      "eval_runtime": 27.2348,
+      "eval_samples_per_second": 18.359,
+      "eval_steps_per_second": 2.313,
       "step": 600
     },
     {
       "epoch": 0.07,
-      "grad_norm": 6.3066606521606445,
+      "grad_norm": 6.283770561218262,
       "learning_rate": 2.3834922328155212e-05,
-      "loss": 1.6176,
+      "loss": 1.6181,
       "step": 700
     },
     {
       "epoch": 0.07,
-      "eval_loss": 1.3781300783157349,
-      "eval_runtime": 27.2227,
-      "eval_samples_per_second": 18.367,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.3765405416488647,
+      "eval_runtime": 27.2404,
+      "eval_samples_per_second": 18.355,
+      "eval_steps_per_second": 2.313,
       "step": 700
     },
     {
       "epoch": 0.08,
-      "grad_norm": 7.34067964553833,
+      "grad_norm": 7.656264781951904,
       "learning_rate": 2.3668244549636642e-05,
-      "loss": 1.628,
+      "loss": 1.6269,
       "step": 800
     },
     {
       "epoch": 0.08,
-      "eval_loss": 1.3644652366638184,
-      "eval_runtime": 27.2246,
-      "eval_samples_per_second": 18.366,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.363603115081787,
+      "eval_runtime": 27.2488,
+      "eval_samples_per_second": 18.349,
+      "eval_steps_per_second": 2.312,
       "step": 800
     },
     {
       "epoch": 0.09,
-      "grad_norm": 4.025296688079834,
+      "grad_norm": 3.9825522899627686,
       "learning_rate": 2.3501566771118076e-05,
-      "loss": 1.5682,
+      "loss": 1.5679,
       "step": 900
     },
     {
       "epoch": 0.09,
-      "eval_loss": 1.3602643013000488,
-      "eval_runtime": 27.2245,
-      "eval_samples_per_second": 18.366,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.3591631650924683,
+      "eval_runtime": 27.2405,
+      "eval_samples_per_second": 18.355,
+      "eval_steps_per_second": 2.313,
       "step": 900
     },
     {
       "epoch": 0.1,
-      "grad_norm": 2.649806499481201,
+      "grad_norm": 2.61617112159729,
       "learning_rate": 2.3334888992599506e-05,
-      "loss": 1.5065,
+      "loss": 1.5055,
       "step": 1000
     },
     {
       "epoch": 0.1,
-      "eval_loss": 1.3477470874786377,
-      "eval_runtime": 27.271,
-      "eval_samples_per_second": 18.334,
-      "eval_steps_per_second": 2.31,
+      "eval_loss": 1.3467729091644287,
+      "eval_runtime": 27.2329,
+      "eval_samples_per_second": 18.36,
+      "eval_steps_per_second": 2.313,
       "step": 1000
     },
     {
       "epoch": 0.11,
-      "grad_norm": 4.1880974769592285,
+      "grad_norm": 4.158337593078613,
       "learning_rate": 2.316821121408094e-05,
-      "loss": 1.5344,
+      "loss": 1.5367,
       "step": 1100
     },
     {
       "epoch": 0.11,
-      "eval_loss": 1.3474398851394653,
-      "eval_runtime": 27.2441,
-      "eval_samples_per_second": 18.353,
-      "eval_steps_per_second": 2.312,
+      "eval_loss": 1.3473803997039795,
+      "eval_runtime": 27.2426,
+      "eval_samples_per_second": 18.354,
+      "eval_steps_per_second": 2.313,
       "step": 1100
     },
     {
       "epoch": 0.12,
-      "grad_norm": 7.770794868469238,
+      "grad_norm": 8.277384757995605,
       "learning_rate": 2.3001533435562373e-05,
-      "loss": 1.5565,
+      "loss": 1.5571,
       "step": 1200
     },
     {
       "epoch": 0.12,
-      "eval_loss": 1.331454873085022,
-      "eval_runtime": 27.2308,
-      "eval_samples_per_second": 18.362,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.3315479755401611,
+      "eval_runtime": 27.2582,
+      "eval_samples_per_second": 18.343,
+      "eval_steps_per_second": 2.311,
       "step": 1200
     },
     {
       "epoch": 0.13,
-      "grad_norm": 17.147653579711914,
+      "grad_norm": 6.899178504943848,
       "learning_rate": 2.2834855657043804e-05,
-      "loss": 1.564,
+      "loss": 1.5624,
       "step": 1300
     },
     {
       "epoch": 0.13,
-      "eval_loss": 1.3280667066574097,
-      "eval_runtime": 27.2409,
-      "eval_samples_per_second": 18.355,
+      "eval_loss": 1.329127311706543,
+      "eval_runtime": 27.2334,
+      "eval_samples_per_second": 18.36,
       "eval_steps_per_second": 2.313,
       "step": 1300
     },
     {
       "epoch": 0.14,
-      "grad_norm": 3.113165855407715,
+      "grad_norm": 3.201148271560669,
       "learning_rate": 2.2668177878525237e-05,
-      "loss": 1.4985,
+      "loss": 1.5005,
       "step": 1400
     },
     {
       "epoch": 0.14,
-      "eval_loss": 1.328255295753479,
-      "eval_runtime": 27.2314,
-      "eval_samples_per_second": 18.361,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.3275610208511353,
+      "eval_runtime": 27.242,
+      "eval_samples_per_second": 18.354,
+      "eval_steps_per_second": 2.313,
       "step": 1400
     },
     {
       "epoch": 0.14,
-      "grad_norm": 8.325815200805664,
+      "grad_norm": 7.74235725402832,
       "learning_rate": 2.2501500100006667e-05,
-      "loss": 1.5745,
+      "loss": 1.5763,
       "step": 1500
     },
     {
       "epoch": 0.14,
-      "eval_loss": 1.3225585222244263,
-      "eval_runtime": 27.2173,
-      "eval_samples_per_second": 18.371,
-      "eval_steps_per_second": 2.315,
+      "eval_loss": 1.3223873376846313,
+      "eval_runtime": 27.2387,
+      "eval_samples_per_second": 18.356,
+      "eval_steps_per_second": 2.313,
       "step": 1500
     },
     {
       "epoch": 0.15,
-      "grad_norm": 2.7448880672454834,
+      "grad_norm": 2.6546003818511963,
       "learning_rate": 2.23348223214881e-05,
-      "loss": 1.4828,
+      "loss": 1.4833,
       "step": 1600
     },
     {
       "epoch": 0.15,
-      "eval_loss": 1.3202852010726929,
-      "eval_runtime": 27.2236,
-      "eval_samples_per_second": 18.366,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.3203272819519043,
+      "eval_runtime": 27.2472,
+      "eval_samples_per_second": 18.35,
+      "eval_steps_per_second": 2.312,
       "step": 1600
     },
     {
       "epoch": 0.16,
-      "grad_norm": 11.465213775634766,
+      "grad_norm": 12.014766693115234,
       "learning_rate": 2.216814454296953e-05,
-      "loss": 1.5091,
+      "loss": 1.5086,
       "step": 1700
     },
     {
       "epoch": 0.16,
-      "eval_loss": 1.3147573471069336,
-      "eval_runtime": 27.2355,
-      "eval_samples_per_second": 18.358,
+      "eval_loss": 1.3142383098602295,
+      "eval_runtime": 27.2319,
+      "eval_samples_per_second": 18.361,
       "eval_steps_per_second": 2.313,
       "step": 1700
     },
     {
       "epoch": 0.17,
-      "grad_norm": 6.742743015289307,
+      "grad_norm": 6.633994102478027,
       "learning_rate": 2.2001466764450965e-05,
-      "loss": 1.5142,
+      "loss": 1.5172,
       "step": 1800
     },
     {
       "epoch": 0.17,
-      "eval_loss": 1.310064435005188,
-      "eval_runtime": 27.2194,
-      "eval_samples_per_second": 18.369,
-      "eval_steps_per_second": 2.315,
+      "eval_loss": 1.3094617128372192,
+      "eval_runtime": 27.2393,
+      "eval_samples_per_second": 18.356,
+      "eval_steps_per_second": 2.313,
       "step": 1800
     },
     {
       "epoch": 0.18,
-      "grad_norm": 6.905991077423096,
+      "grad_norm": 6.995003700256348,
       "learning_rate": 2.18347889859324e-05,
-      "loss": 1.4293,
+      "loss": 1.4299,
       "step": 1900
     },
     {
       "epoch": 0.18,
-      "eval_loss": 1.303736686706543,
-      "eval_runtime": 27.2248,
-      "eval_samples_per_second": 18.366,
+      "eval_loss": 1.3035434484481812,
+      "eval_runtime": 27.2303,
+      "eval_samples_per_second": 18.362,
       "eval_steps_per_second": 2.314,
       "step": 1900
     },
     {
       "epoch": 0.19,
-      "grad_norm": 7.1159162521362305,
+      "grad_norm": 6.952856540679932,
       "learning_rate": 2.166811120741383e-05,
-      "loss": 1.6055,
+      "loss": 1.6068,
       "step": 2000
     },
     {
       "epoch": 0.19,
-      "eval_loss": 1.2934476137161255,
-      "eval_runtime": 27.2203,
-      "eval_samples_per_second": 18.369,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.2929936647415161,
+      "eval_runtime": 27.2469,
+      "eval_samples_per_second": 18.351,
+      "eval_steps_per_second": 2.312,
       "step": 2000
     },
     {
       "epoch": 0.2,
-      "grad_norm": 7.020285606384277,
+      "grad_norm": 7.213510036468506,
       "learning_rate": 2.150143342889526e-05,
-      "loss": 1.526,
+      "loss": 1.5253,
       "step": 2100
     },
     {
       "epoch": 0.2,
-      "eval_loss": 1.295276165008545,
-      "eval_runtime": 27.2359,
-      "eval_samples_per_second": 18.358,
+      "eval_loss": 1.294145941734314,
+      "eval_runtime": 27.2323,
+      "eval_samples_per_second": 18.361,
       "eval_steps_per_second": 2.313,
       "step": 2100
     },
     {
       "epoch": 0.21,
-      "grad_norm": 12.202252388000488,
+      "grad_norm": 10.12060832977295,
       "learning_rate": 2.1334755650376693e-05,
-      "loss": 1.5421,
+      "loss": 1.54,
       "step": 2200
     },
     {
       "epoch": 0.21,
-      "eval_loss": 1.2898857593536377,
-      "eval_runtime": 27.2174,
-      "eval_samples_per_second": 18.371,
-      "eval_steps_per_second": 2.315,
+      "eval_loss": 1.2877910137176514,
+      "eval_runtime": 27.2416,
+      "eval_samples_per_second": 18.354,
+      "eval_steps_per_second": 2.313,
       "step": 2200
     },
     {
       "epoch": 0.22,
-      "grad_norm": 8.110170364379883,
+      "grad_norm": 9.764932632446289,
       "learning_rate": 2.1168077871858126e-05,
-      "loss": 1.5439,
+      "loss": 1.5397,
       "step": 2300
     },
     {
       "epoch": 0.22,
-      "eval_loss": 1.2835201025009155,
-      "eval_runtime": 27.234,
-      "eval_samples_per_second": 18.359,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.2824313640594482,
+      "eval_runtime": 27.2505,
+      "eval_samples_per_second": 18.348,
+      "eval_steps_per_second": 2.312,
       "step": 2300
     },
     {
       "epoch": 0.23,
-      "grad_norm": 9.563570022583008,
+      "grad_norm": 7.659645080566406,
       "learning_rate": 2.1001400093339556e-05,
-      "loss": 1.4597,
+      "loss": 1.4568,
       "step": 2400
     },
     {
       "epoch": 0.23,
-      "eval_loss": 1.2785178422927856,
-      "eval_runtime": 27.2163,
-      "eval_samples_per_second": 18.371,
-      "eval_steps_per_second": 2.315,
+      "eval_loss": 1.277322769165039,
+      "eval_runtime": 27.244,
+      "eval_samples_per_second": 18.353,
+      "eval_steps_per_second": 2.312,
       "step": 2400
     },
     {
       "epoch": 0.24,
-      "grad_norm": 14.292014122009277,
+      "grad_norm": 9.128131866455078,
       "learning_rate": 2.083472231482099e-05,
-      "loss": 1.4039,
+      "loss": 1.4044,
       "step": 2500
     },
     {
       "epoch": 0.24,
-      "eval_loss": 1.2762665748596191,
-      "eval_runtime": 27.2214,
-      "eval_samples_per_second": 18.368,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.274811863899231,
+      "eval_runtime": 27.2388,
+      "eval_samples_per_second": 18.356,
+      "eval_steps_per_second": 2.313,
       "step": 2500
     },
     {
       "epoch": 0.25,
-      "grad_norm": 4.076394081115723,
+      "grad_norm": 3.9920785427093506,
       "learning_rate": 2.0668044536302424e-05,
-      "loss": 1.5642,
+      "loss": 1.5648,
       "step": 2600
     },
     {
       "epoch": 0.25,
-      "eval_loss": 1.2759555578231812,
-      "eval_runtime": 27.2363,
-      "eval_samples_per_second": 18.358,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.2758684158325195,
+      "eval_runtime": 27.2442,
+      "eval_samples_per_second": 18.353,
+      "eval_steps_per_second": 2.312,
       "step": 2600
     },
     {
       "epoch": 0.26,
-      "grad_norm": 10.41948413848877,
+      "grad_norm": 10.10120677947998,
       "learning_rate": 2.050136675778385e-05,
-      "loss": 1.3917,
+      "loss": 1.3923,
       "step": 2700
     },
     {
       "epoch": 0.26,
-      "eval_loss": 1.2696521282196045,
-      "eval_runtime": 27.2277,
-      "eval_samples_per_second": 18.364,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.2696377038955688,
+      "eval_runtime": 27.2457,
+      "eval_samples_per_second": 18.352,
+      "eval_steps_per_second": 2.312,
       "step": 2700
     },
     {
       "epoch": 0.27,
-      "grad_norm": 8.529683113098145,
+      "grad_norm": 8.736724853515625,
       "learning_rate": 2.0334688979265284e-05,
-      "loss": 1.5147,
+      "loss": 1.5071,
       "step": 2800
     },
     {
       "epoch": 0.27,
-      "eval_loss": 1.2705649137496948,
-      "eval_runtime": 27.2409,
-      "eval_samples_per_second": 18.355,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.269506812095642,
+      "eval_runtime": 27.2434,
+      "eval_samples_per_second": 18.353,
+      "eval_steps_per_second": 2.312,
       "step": 2800
     },
     {
       "epoch": 0.28,
-      "grad_norm": 4.257082939147949,
+      "grad_norm": 4.389509677886963,
       "learning_rate": 2.0168011200746718e-05,
-      "loss": 1.4392,
+      "loss": 1.4408,
       "step": 2900
     },
     {
       "epoch": 0.28,
-      "eval_loss": 1.2686102390289307,
-      "eval_runtime": 27.2394,
-      "eval_samples_per_second": 18.356,
+      "eval_loss": 1.2669789791107178,
+      "eval_runtime": 27.2401,
+      "eval_samples_per_second": 18.355,
       "eval_steps_per_second": 2.313,
       "step": 2900
     },
     {
       "epoch": 0.29,
-      "grad_norm": 9.243705749511719,
+      "grad_norm": 7.288300514221191,
       "learning_rate": 2.0001333422228148e-05,
-      "loss": 1.42,
+      "loss": 1.4174,
       "step": 3000
     },
     {
       "epoch": 0.29,
-      "eval_loss": 1.2711559534072876,
-      "eval_runtime": 27.2345,
-      "eval_samples_per_second": 18.359,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.2704719305038452,
+      "eval_runtime": 27.225,
+      "eval_samples_per_second": 18.365,
+      "eval_steps_per_second": 2.314,
       "step": 3000
     },
     {
       "epoch": 0.3,
-      "grad_norm": 3.96095609664917,
+      "grad_norm": 3.918144464492798,
       "learning_rate": 1.983465564370958e-05,
-      "loss": 1.4526,
+      "loss": 1.4545,
       "step": 3100
     },
     {
       "epoch": 0.3,
-      "eval_loss": 1.2658092975616455,
-      "eval_runtime": 27.251,
-      "eval_samples_per_second": 18.348,
-      "eval_steps_per_second": 2.312,
+      "eval_loss": 1.2649730443954468,
+      "eval_runtime": 27.2304,
+      "eval_samples_per_second": 18.362,
+      "eval_steps_per_second": 2.314,
       "step": 3100
     },
     {
       "epoch": 0.31,
-      "grad_norm": 7.505685329437256,
+      "grad_norm": 7.6044392585754395,
       "learning_rate": 1.9667977865191015e-05,
-      "loss": 1.464,
+      "loss": 1.4661,
       "step": 3200
     },
     {
       "epoch": 0.31,
-      "eval_loss": 1.2660542726516724,
-      "eval_runtime": 27.2428,
-      "eval_samples_per_second": 18.354,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.2659434080123901,
+      "eval_runtime": 27.2453,
+      "eval_samples_per_second": 18.352,
+      "eval_steps_per_second": 2.312,
       "step": 3200
     },
     {
       "epoch": 0.32,
-      "grad_norm": 2.5489604473114014,
+      "grad_norm": 2.494729518890381,
       "learning_rate": 1.9501300086672446e-05,
-      "loss": 1.4093,
+      "loss": 1.409,
       "step": 3300
     },
     {
       "epoch": 0.32,
-      "eval_loss": 1.2600222826004028,
-      "eval_runtime": 27.2413,
-      "eval_samples_per_second": 18.355,
+      "eval_loss": 1.2590056657791138,
+      "eval_runtime": 27.2367,
+      "eval_samples_per_second": 18.358,
       "eval_steps_per_second": 2.313,
       "step": 3300
     },
     {
       "epoch": 0.33,
-      "grad_norm": 5.240612506866455,
+      "grad_norm": 5.691107273101807,
       "learning_rate": 1.9334622308153876e-05,
-      "loss": 1.4735,
+      "loss": 1.4775,
       "step": 3400
     },
     {
       "epoch": 0.33,
-      "eval_loss": 1.2596392631530762,
-      "eval_runtime": 27.2353,
-      "eval_samples_per_second": 18.359,
+      "eval_loss": 1.2585530281066895,
+      "eval_runtime": 27.2332,
+      "eval_samples_per_second": 18.36,
       "eval_steps_per_second": 2.313,
       "step": 3400
     },
     {
       "epoch": 0.34,
-      "grad_norm": 7.678092002868652,
+      "grad_norm": 8.363253593444824,
       "learning_rate": 1.916794452963531e-05,
-      "loss": 1.377,
+      "loss": 1.3749,
       "step": 3500
     },
     {
       "epoch": 0.34,
-      "eval_loss": 1.2602125406265259,
-      "eval_runtime": 27.2377,
-      "eval_samples_per_second": 18.357,
+      "eval_loss": 1.2596063613891602,
+      "eval_runtime": 27.2327,
+      "eval_samples_per_second": 18.36,
       "eval_steps_per_second": 2.313,
       "step": 3500
     },
     {
       "epoch": 0.35,
-      "grad_norm": 5.745735168457031,
+      "grad_norm": 5.728103160858154,
       "learning_rate": 1.9001266751116743e-05,
-      "loss": 1.4398,
+      "loss": 1.438,
       "step": 3600
     },
     {
       "epoch": 0.35,
-      "eval_loss": 1.2613269090652466,
-      "eval_runtime": 27.2388,
-      "eval_samples_per_second": 18.356,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.25990629196167,
+      "eval_runtime": 27.2309,
+      "eval_samples_per_second": 18.361,
+      "eval_steps_per_second": 2.314,
       "step": 3600
     },
     {
       "epoch": 0.36,
-      "grad_norm": 6.888312816619873,
+      "grad_norm": 8.10669994354248,
       "learning_rate": 1.8834588972598173e-05,
-      "loss": 1.4953,
+      "loss": 1.4927,
       "step": 3700
     },
     {
       "epoch": 0.36,
-      "eval_loss": 1.2581281661987305,
-      "eval_runtime": 27.2396,
-      "eval_samples_per_second": 18.356,
+      "eval_loss": 1.2571576833724976,
+      "eval_runtime": 27.2321,
+      "eval_samples_per_second": 18.361,
       "eval_steps_per_second": 2.313,
       "step": 3700
     },
     {
       "epoch": 0.37,
-      "grad_norm": 6.642201900482178,
+      "grad_norm": 5.870044231414795,
       "learning_rate": 1.8667911194079607e-05,
-      "loss": 1.3507,
+      "loss": 1.3475,
       "step": 3800
     },
     {
       "epoch": 0.37,
-      "eval_loss": 1.2513525485992432,
-      "eval_runtime": 27.2312,
-      "eval_samples_per_second": 18.361,
+      "eval_loss": 1.2502912282943726,
+      "eval_runtime": 27.2283,
+      "eval_samples_per_second": 18.363,
       "eval_steps_per_second": 2.314,
       "step": 3800
     },
     {
       "epoch": 0.38,
-      "grad_norm": 8.896329879760742,
+      "grad_norm": 8.73828125,
       "learning_rate": 1.850123341556104e-05,
-      "loss": 1.4659,
+      "loss": 1.467,
       "step": 3900
     },
     {
       "epoch": 0.38,
-      "eval_loss": 1.252593994140625,
-      "eval_runtime": 27.2402,
-      "eval_samples_per_second": 18.355,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.250830054283142,
+      "eval_runtime": 27.2279,
+      "eval_samples_per_second": 18.364,
+      "eval_steps_per_second": 2.314,
       "step": 3900
     },
     {
       "epoch": 0.39,
-      "grad_norm": 6.306306838989258,
+      "grad_norm": 6.171374797821045,
       "learning_rate": 1.833455563704247e-05,
-      "loss": 1.3839,
+      "loss": 1.3805,
       "step": 4000
     },
     {
       "epoch": 0.39,
-      "eval_loss": 1.2493339776992798,
-      "eval_runtime": 27.2431,
-      "eval_samples_per_second": 18.353,
+      "eval_loss": 1.2475095987319946,
+      "eval_runtime": 27.2405,
+      "eval_samples_per_second": 18.355,
       "eval_steps_per_second": 2.313,
       "step": 4000
     },
     {
       "epoch": 0.4,
-      "grad_norm": 9.476497650146484,
+      "grad_norm": 9.766369819641113,
       "learning_rate": 1.81678778585239e-05,
-      "loss": 1.4575,
+      "loss": 1.4581,
       "step": 4100
     },
     {
       "epoch": 0.4,
-      "eval_loss": 1.2447402477264404,
-      "eval_runtime": 27.2408,
-      "eval_samples_per_second": 18.355,
+      "eval_loss": 1.2426196336746216,
+      "eval_runtime": 27.2316,
+      "eval_samples_per_second": 18.361,
       "eval_steps_per_second": 2.313,
       "step": 4100
     },
     {
       "epoch": 0.41,
-      "grad_norm": 8.788684844970703,
+      "grad_norm": 8.565445899963379,
       "learning_rate": 1.8001200080005335e-05,
-      "loss": 1.4355,
+      "loss": 1.4319,
       "step": 4200
     },
     {
       "epoch": 0.41,
-      "eval_loss": 1.2466756105422974,
-      "eval_runtime": 27.2362,
-      "eval_samples_per_second": 18.358,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.2432831525802612,
+      "eval_runtime": 27.2265,
+      "eval_samples_per_second": 18.364,
+      "eval_steps_per_second": 2.314,
       "step": 4200
     },
     {
       "epoch": 0.42,
-      "grad_norm": 7.325284004211426,
+      "grad_norm": 7.835436820983887,
       "learning_rate": 1.7834522301486768e-05,
-      "loss": 1.4325,
+      "loss": 1.4282,
       "step": 4300
     },
     {
       "epoch": 0.42,
-      "eval_loss": 1.2401726245880127,
-      "eval_runtime": 27.2408,
-      "eval_samples_per_second": 18.355,
+      "eval_loss": 1.23819899559021,
+      "eval_runtime": 27.239,
+      "eval_samples_per_second": 18.356,
       "eval_steps_per_second": 2.313,
       "step": 4300
     },
     {
       "epoch": 0.42,
-      "grad_norm": 5.760813236236572,
+      "grad_norm": 5.4964985847473145,
       "learning_rate": 1.76678445229682e-05,
-      "loss": 1.444,
+      "loss": 1.4429,
       "step": 4400
     },
     {
       "epoch": 0.42,
-      "eval_loss": 1.237419843673706,
-      "eval_runtime": 27.235,
-      "eval_samples_per_second": 18.359,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.2355051040649414,
+      "eval_runtime": 27.2454,
+      "eval_samples_per_second": 18.352,
+      "eval_steps_per_second": 2.312,
       "step": 4400
     },
     {
       "epoch": 0.43,
-      "grad_norm": 5.730567932128906,
+      "grad_norm": 6.0724897384643555,
       "learning_rate": 1.7501166744449632e-05,
-      "loss": 1.3808,
+      "loss": 1.3815,
       "step": 4500
     },
     {
       "epoch": 0.43,
-      "eval_loss": 1.233359932899475,
-      "eval_runtime": 27.2413,
-      "eval_samples_per_second": 18.355,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.2307285070419312,
+      "eval_runtime": 27.248,
+      "eval_samples_per_second": 18.35,
+      "eval_steps_per_second": 2.312,
       "step": 4500
     },
     {
       "epoch": 0.44,
-      "grad_norm": 5.33050537109375,
+      "grad_norm": 5.445316314697266,
       "learning_rate": 1.7334488965931062e-05,
-      "loss": 1.4183,
+      "loss": 1.4194,
       "step": 4600
     },
     {
       "epoch": 0.44,
-      "eval_loss": 1.2271193265914917,
-      "eval_runtime": 27.2354,
-      "eval_samples_per_second": 18.358,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.225948691368103,
+      "eval_runtime": 27.2592,
+      "eval_samples_per_second": 18.342,
+      "eval_steps_per_second": 2.311,
       "step": 4600
     },
     {
       "epoch": 0.45,
-      "grad_norm": 5.542807102203369,
+      "grad_norm": 6.078861236572266,
       "learning_rate": 1.7167811187412493e-05,
-      "loss": 1.4737,
+      "loss": 1.4719,
       "step": 4700
     },
     {
       "epoch": 0.45,
-      "eval_loss": 1.2270761728286743,
-      "eval_runtime": 27.2428,
-      "eval_samples_per_second": 18.353,
+      "eval_loss": 1.2257155179977417,
+      "eval_runtime": 27.234,
+      "eval_samples_per_second": 18.359,
       "eval_steps_per_second": 2.313,
       "step": 4700
     },
     {
       "epoch": 0.46,
-      "grad_norm": 5.02695369720459,
+      "grad_norm": 4.971687316894531,
       "learning_rate": 1.7001133408893926e-05,
-      "loss": 1.3564,
+      "loss": 1.3587,
       "step": 4800
     },
     {
       "epoch": 0.46,
-      "eval_loss": 1.2189486026763916,
-      "eval_runtime": 27.2289,
-      "eval_samples_per_second": 18.363,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.219119668006897,
+      "eval_runtime": 27.232,
+      "eval_samples_per_second": 18.361,
+      "eval_steps_per_second": 2.313,
       "step": 4800
     },
     {
       "epoch": 0.47,
-      "grad_norm": 9.54018783569336,
+      "grad_norm": 10.197978973388672,
       "learning_rate": 1.683445563037536e-05,
-      "loss": 1.3615,
+      "loss": 1.3599,
       "step": 4900
     },
     {
       "epoch": 0.47,
-      "eval_loss": 1.2135944366455078,
-      "eval_runtime": 27.2481,
-      "eval_samples_per_second": 18.35,
-      "eval_steps_per_second": 2.312,
+      "eval_loss": 1.2111576795578003,
+      "eval_runtime": 27.2374,
+      "eval_samples_per_second": 18.357,
+      "eval_steps_per_second": 2.313,
       "step": 4900
     },
     {
       "epoch": 0.48,
-      "grad_norm": 4.637027740478516,
+      "grad_norm": 4.264425754547119,
       "learning_rate": 1.6667777851856793e-05,
-      "loss": 1.4264,
+      "loss": 1.4247,
       "step": 5000
     },
     {
       "epoch": 0.48,
-      "eval_loss": 1.2105356454849243,
-      "eval_runtime": 27.2429,
-      "eval_samples_per_second": 18.353,
+      "eval_loss": 1.207905650138855,
+      "eval_runtime": 27.2335,
+      "eval_samples_per_second": 18.36,
       "eval_steps_per_second": 2.313,
       "step": 5000
     },
     {
       "epoch": 0.49,
-      "grad_norm": 7.059263706207275,
+      "grad_norm": 6.512348651885986,
       "learning_rate": 1.6501100073338224e-05,
-      "loss": 1.281,
+      "loss": 1.2745,
       "step": 5100
     },
     {
       "epoch": 0.49,
-      "eval_loss": 1.2117621898651123,
-      "eval_runtime": 27.2335,
-      "eval_samples_per_second": 18.36,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.210147738456726,
+      "eval_runtime": 27.2312,
+      "eval_samples_per_second": 18.361,
+      "eval_steps_per_second": 2.314,
       "step": 5100
     },
     {
       "epoch": 0.5,
-      "grad_norm": 4.905197620391846,
+      "grad_norm": 5.255068778991699,
       "learning_rate": 1.6334422294819657e-05,
-      "loss": 1.4319,
+      "loss": 1.4317,
       "step": 5200
     },
     {
       "epoch": 0.5,
-      "eval_loss": 1.2079901695251465,
-      "eval_runtime": 27.2435,
-      "eval_samples_per_second": 18.353,
-      "eval_steps_per_second": 2.312,
+      "eval_loss": 1.2059623003005981,
+      "eval_runtime": 27.2372,
+      "eval_samples_per_second": 18.357,
+      "eval_steps_per_second": 2.313,
       "step": 5200
     },
     {
       "epoch": 0.51,
-      "grad_norm": 11.458580017089844,
+      "grad_norm": 11.69495677947998,
       "learning_rate": 1.6167744516301088e-05,
-      "loss": 1.346,
+      "loss": 1.3463,
       "step": 5300
     },
     {
       "epoch": 0.51,
-      "eval_loss": 1.2106558084487915,
-      "eval_runtime": 27.2322,
-      "eval_samples_per_second": 18.361,
+      "eval_loss": 1.2085806131362915,
+      "eval_runtime": 27.238,
+      "eval_samples_per_second": 18.357,
       "eval_steps_per_second": 2.313,
       "step": 5300
     },
     {
       "epoch": 0.52,
-      "grad_norm": 5.2320146560668945,
+      "grad_norm": 4.8801398277282715,
       "learning_rate": 1.6001066737782518e-05,
-      "loss": 1.396,
+      "loss": 1.3995,
       "step": 5400
     },
     {
       "epoch": 0.52,
-      "eval_loss": 1.208406925201416,
-      "eval_runtime": 27.2304,
-      "eval_samples_per_second": 18.362,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.2063578367233276,
+      "eval_runtime": 27.2504,
+      "eval_samples_per_second": 18.348,
+      "eval_steps_per_second": 2.312,
       "step": 5400
     },
     {
       "epoch": 0.53,
-      "grad_norm": 7.270910739898682,
+      "grad_norm": 6.397332668304443,
       "learning_rate": 1.583438895926395e-05,
-      "loss": 1.3313,
+      "loss": 1.3316,
       "step": 5500
     },
     {
       "epoch": 0.53,
-      "eval_loss": 1.2069458961486816,
-      "eval_runtime": 27.2355,
-      "eval_samples_per_second": 18.358,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.2056635618209839,
+      "eval_runtime": 27.2573,
+      "eval_samples_per_second": 18.344,
+      "eval_steps_per_second": 2.311,
       "step": 5500
     },
     {
       "epoch": 0.54,
-      "grad_norm": 8.65515422821045,
+      "grad_norm": 8.16231918334961,
       "learning_rate": 1.5667711180745385e-05,
-      "loss": 1.3069,
+      "loss": 1.3064,
       "step": 5600
     },
     {
       "epoch": 0.54,
-      "eval_loss": 1.2003055810928345,
-      "eval_runtime": 27.2389,
-      "eval_samples_per_second": 18.356,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.1981565952301025,
+      "eval_runtime": 27.2447,
+      "eval_samples_per_second": 18.352,
+      "eval_steps_per_second": 2.312,
       "step": 5600
     },
     {
       "epoch": 0.55,
-      "grad_norm": 7.7110724449157715,
+      "grad_norm": 8.272992134094238,
       "learning_rate": 1.5501033402226815e-05,
-      "loss": 1.2984,
+      "loss": 1.2971,
       "step": 5700
     },
     {
       "epoch": 0.55,
-      "eval_loss": 1.20066499710083,
-      "eval_runtime": 27.2419,
-      "eval_samples_per_second": 18.354,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.1983332633972168,
+      "eval_runtime": 27.2527,
+      "eval_samples_per_second": 18.347,
+      "eval_steps_per_second": 2.312,
       "step": 5700
     },
     {
       "epoch": 0.56,
-      "grad_norm": 7.086616516113281,
+      "grad_norm": 7.432589054107666,
       "learning_rate": 1.533435562370825e-05,
-      "loss": 1.3998,
+      "loss": 1.4006,
       "step": 5800
     },
     {
       "epoch": 0.56,
-      "eval_loss": 1.1993930339813232,
-      "eval_runtime": 27.2542,
-      "eval_samples_per_second": 18.346,
-      "eval_steps_per_second": 2.312,
+      "eval_loss": 1.1970897912979126,
+      "eval_runtime": 27.2266,
+      "eval_samples_per_second": 18.364,
+      "eval_steps_per_second": 2.314,
       "step": 5800
     },
     {
       "epoch": 0.57,
-      "grad_norm": 8.898906707763672,
+      "grad_norm": 12.17546272277832,
       "learning_rate": 1.5167677845189679e-05,
-      "loss": 1.4083,
+      "loss": 1.4033,
       "step": 5900
     },
     {
       "epoch": 0.57,
-      "eval_loss": 1.2009773254394531,
-      "eval_runtime": 27.2476,
+      "eval_loss": 1.1994664669036865,
+      "eval_runtime": 27.2478,
       "eval_samples_per_second": 18.35,
       "eval_steps_per_second": 2.312,
       "step": 5900
     },
     {
       "epoch": 0.58,
-      "grad_norm": 6.958111763000488,
+      "grad_norm": 6.728982448577881,
       "learning_rate": 1.5001000066671111e-05,
-      "loss": 1.3425,
+      "loss": 1.3381,
       "step": 6000
     },
     {
       "epoch": 0.58,
-      "eval_loss": 1.1988615989685059,
-      "eval_runtime": 27.2347,
+      "eval_loss": 1.197486162185669,
+      "eval_runtime": 27.234,
       "eval_samples_per_second": 18.359,
       "eval_steps_per_second": 2.313,
       "step": 6000
     },
     {
       "epoch": 0.59,
-      "grad_norm": 5.355065822601318,
+      "grad_norm": 5.387950420379639,
       "learning_rate": 1.4834322288152545e-05,
-      "loss": 1.3308,
+      "loss": 1.3284,
       "step": 6100
     },
     {
       "epoch": 0.59,
-      "eval_loss": 1.1959185600280762,
-      "eval_runtime": 27.2351,
-      "eval_samples_per_second": 18.359,
+      "eval_loss": 1.19466233253479,
+      "eval_runtime": 27.2424,
+      "eval_samples_per_second": 18.354,
       "eval_steps_per_second": 2.313,
       "step": 6100
     },
     {
       "epoch": 0.6,
-      "grad_norm": 6.108639240264893,
+      "grad_norm": 6.059466361999512,
       "learning_rate": 1.4667644509633977e-05,
-      "loss": 1.3496,
+      "loss": 1.3492,
       "step": 6200
     },
     {
       "epoch": 0.6,
-      "eval_loss": 1.1931555271148682,
-      "eval_runtime": 27.2299,
-      "eval_samples_per_second": 18.362,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.191383719444275,
+      "eval_runtime": 27.2468,
+      "eval_samples_per_second": 18.351,
+      "eval_steps_per_second": 2.312,
       "step": 6200
     },
     {
       "epoch": 0.61,
-      "grad_norm": 5.120665550231934,
+      "grad_norm": 4.771157264709473,
       "learning_rate": 1.4500966731115409e-05,
-      "loss": 1.2974,
+      "loss": 1.2956,
       "step": 6300
     },
     {
       "epoch": 0.61,
-      "eval_loss": 1.189644455909729,
-      "eval_runtime": 27.2414,
-      "eval_samples_per_second": 18.354,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.1882104873657227,
+      "eval_runtime": 27.2582,
+      "eval_samples_per_second": 18.343,
+      "eval_steps_per_second": 2.311,
       "step": 6300
     },
     {
       "epoch": 0.62,
-      "grad_norm": 9.152058601379395,
+      "grad_norm": 8.666051864624023,
       "learning_rate": 1.4334288952596842e-05,
-      "loss": 1.3366,
+      "loss": 1.3381,
       "step": 6400
     },
     {
       "epoch": 0.62,
-      "eval_loss": 1.1901721954345703,
-      "eval_runtime": 27.2454,
-      "eval_samples_per_second": 18.352,
-      "eval_steps_per_second": 2.312,
+      "eval_loss": 1.1888271570205688,
+      "eval_runtime": 27.2369,
+      "eval_samples_per_second": 18.357,
+      "eval_steps_per_second": 2.313,
       "step": 6400
     },
     {
       "epoch": 0.63,
-      "grad_norm": 4.623256683349609,
+      "grad_norm": 4.501246452331543,
       "learning_rate": 1.416761117407827e-05,
-      "loss": 1.4328,
+      "loss": 1.4293,
       "step": 6500
     },
     {
       "epoch": 0.63,
-      "eval_loss": 1.1861129999160767,
-      "eval_runtime": 27.2464,
-      "eval_samples_per_second": 18.351,
-      "eval_steps_per_second": 2.312,
+      "eval_loss": 1.1847625970840454,
+      "eval_runtime": 27.239,
+      "eval_samples_per_second": 18.356,
+      "eval_steps_per_second": 2.313,
       "step": 6500
     },
     {
       "epoch": 0.64,
-      "grad_norm": 5.927731037139893,
+      "grad_norm": 7.15999174118042,
       "learning_rate": 1.4000933395559704e-05,
-      "loss": 1.2727,
+      "loss": 1.2713,
       "step": 6600
     },
     {
       "epoch": 0.64,
-      "eval_loss": 1.1859809160232544,
-      "eval_runtime": 27.2453,
-      "eval_samples_per_second": 18.352,
+      "eval_loss": 1.1845014095306396,
+      "eval_runtime": 27.2437,
+      "eval_samples_per_second": 18.353,
       "eval_steps_per_second": 2.312,
       "step": 6600
     },
     {
       "epoch": 0.65,
-      "grad_norm": 4.167983055114746,
+      "grad_norm": 4.205693244934082,
       "learning_rate": 1.3834255617041136e-05,
-      "loss": 1.405,
+      "loss": 1.4094,
       "step": 6700
     },
     {
       "epoch": 0.65,
-      "eval_loss": 1.1869107484817505,
-      "eval_runtime": 27.2415,
-      "eval_samples_per_second": 18.354,
+      "eval_loss": 1.1849124431610107,
+      "eval_runtime": 27.2322,
+      "eval_samples_per_second": 18.361,
       "eval_steps_per_second": 2.313,
       "step": 6700
     },
     {
       "epoch": 0.66,
-      "grad_norm": 11.508980751037598,
+      "grad_norm": 10.90744400024414,
       "learning_rate": 1.3667577838522568e-05,
-      "loss": 1.3313,
+      "loss": 1.3321,
       "step": 6800
     },
     {
       "epoch": 0.66,
-      "eval_loss": 1.1844249963760376,
-      "eval_runtime": 27.2335,
-      "eval_samples_per_second": 18.36,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.1828891038894653,
+      "eval_runtime": 27.2545,
+      "eval_samples_per_second": 18.346,
+      "eval_steps_per_second": 2.312,
       "step": 6800
     },
     {
       "epoch": 0.67,
-      "grad_norm": 5.97917366027832,
+      "grad_norm": 5.98179292678833,
       "learning_rate": 1.3500900060004002e-05,
-      "loss": 1.3411,
+      "loss": 1.3464,
       "step": 6900
     },
     {
       "epoch": 0.67,
-      "eval_loss": 1.1828359365463257,
-      "eval_runtime": 27.2322,
-      "eval_samples_per_second": 18.361,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.181032419204712,
+      "eval_runtime": 27.2781,
+      "eval_samples_per_second": 18.33,
+      "eval_steps_per_second": 2.31,
       "step": 6900
     },
     {
       "epoch": 0.68,
-      "grad_norm": 7.086197376251221,
+      "grad_norm": 7.574715614318848,
       "learning_rate": 1.3334222281485434e-05,
-      "loss": 1.3896,
+      "loss": 1.3912,
       "step": 7000
     },
     {
       "epoch": 0.68,
-      "eval_loss": 1.1808134317398071,
-      "eval_runtime": 27.2433,
-      "eval_samples_per_second": 18.353,
-      "eval_steps_per_second": 2.312,
+      "eval_loss": 1.178390383720398,
+      "eval_runtime": 27.2322,
+      "eval_samples_per_second": 18.361,
+      "eval_steps_per_second": 2.313,
       "step": 7000
     },
     {
       "epoch": 0.69,
-      "grad_norm": 7.553659439086914,
+      "grad_norm": 7.541271209716797,
       "learning_rate": 1.3167544502966864e-05,
-      "loss": 1.3611,
+      "loss": 1.3588,
       "step": 7100
     },
     {
       "epoch": 0.69,
-      "eval_loss": 1.1791377067565918,
-      "eval_runtime": 27.2454,
+      "eval_loss": 1.177061915397644,
+      "eval_runtime": 27.2447,
       "eval_samples_per_second": 18.352,
       "eval_steps_per_second": 2.312,
       "step": 7100
     },
     {
       "epoch": 0.69,
-      "grad_norm": 8.123412132263184,
+      "grad_norm": 9.224418640136719,
       "learning_rate": 1.3000866724448296e-05,
       "loss": 1.3083,
       "step": 7200
     },
     {
       "epoch": 0.69,
-      "eval_loss": 1.177075982093811,
-      "eval_runtime": 27.2307,
-      "eval_samples_per_second": 18.362,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.1750773191452026,
+      "eval_runtime": 27.2423,
+      "eval_samples_per_second": 18.354,
+      "eval_steps_per_second": 2.313,
       "step": 7200
     },
     {
       "epoch": 0.7,
-      "grad_norm": 9.786763191223145,
+      "grad_norm": 9.978618621826172,
       "learning_rate": 1.283418894592973e-05,
-      "loss": 1.2965,
+      "loss": 1.2933,
       "step": 7300
     },
     {
       "epoch": 0.7,
-      "eval_loss": 1.1764291524887085,
-      "eval_runtime": 27.2358,
-      "eval_samples_per_second": 18.358,
+      "eval_loss": 1.174636960029602,
+      "eval_runtime": 27.2335,
+      "eval_samples_per_second": 18.36,
       "eval_steps_per_second": 2.313,
       "step": 7300
     },
     {
       "epoch": 0.71,
-      "grad_norm": 6.950915336608887,
+      "grad_norm": 6.905595779418945,
       "learning_rate": 1.2667511167411161e-05,
-      "loss": 1.3109,
+      "loss": 1.3069,
       "step": 7400
     },
     {
       "epoch": 0.71,
-      "eval_loss": 1.1759792566299438,
-      "eval_runtime": 27.2377,
-      "eval_samples_per_second": 18.357,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.1744636297225952,
+      "eval_runtime": 27.2307,
+      "eval_samples_per_second": 18.362,
+      "eval_steps_per_second": 2.314,
       "step": 7400
     },
     {
       "epoch": 0.72,
-      "grad_norm": 6.95255184173584,
+      "grad_norm": 7.499482154846191,
       "learning_rate": 1.2500833388892593e-05,
-      "loss": 1.317,
+      "loss": 1.3172,
       "step": 7500
     },
     {
       "epoch": 0.72,
-      "eval_loss": 1.1741067171096802,
-      "eval_runtime": 27.2348,
-      "eval_samples_per_second": 18.359,
+      "eval_loss": 1.173417568206787,
+      "eval_runtime": 27.2368,
+      "eval_samples_per_second": 18.357,
       "eval_steps_per_second": 2.313,
       "step": 7500
     },
     {
       "epoch": 0.73,
-      "grad_norm": 4.61511754989624,
+      "grad_norm": 4.5594282150268555,
       "learning_rate": 1.2334155610374025e-05,
-      "loss": 1.3408,
+      "loss": 1.3392,
       "step": 7600
     },
     {
       "epoch": 0.73,
-      "eval_loss": 1.1740319728851318,
-      "eval_runtime": 27.2347,
-      "eval_samples_per_second": 18.359,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.1731162071228027,
+      "eval_runtime": 27.2663,
+      "eval_samples_per_second": 18.338,
+      "eval_steps_per_second": 2.311,
       "step": 7600
     },
     {
       "epoch": 0.74,
-      "grad_norm": 4.9650750160217285,
+      "grad_norm": 4.877806663513184,
       "learning_rate": 1.2167477831855457e-05,
-      "loss": 1.2952,
+      "loss": 1.2941,
       "step": 7700
     },
     {
       "epoch": 0.74,
-      "eval_loss": 1.1764243841171265,
-      "eval_runtime": 27.2348,
-      "eval_samples_per_second": 18.359,
+      "eval_loss": 1.1754953861236572,
+      "eval_runtime": 27.2391,
+      "eval_samples_per_second": 18.356,
       "eval_steps_per_second": 2.313,
       "step": 7700
     },
     {
       "epoch": 0.75,
-      "grad_norm": 7.219305515289307,
+      "grad_norm": 6.909502029418945,
       "learning_rate": 1.200080005333689e-05,
-      "loss": 1.302,
+      "loss": 1.3124,
       "step": 7800
     },
     {
       "epoch": 0.75,
-      "eval_loss": 1.1715896129608154,
-      "eval_runtime": 27.2301,
-      "eval_samples_per_second": 18.362,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.1714600324630737,
+      "eval_runtime": 27.2351,
+      "eval_samples_per_second": 18.359,
+      "eval_steps_per_second": 2.313,
       "step": 7800
     },
     {
       "epoch": 0.76,
-      "grad_norm": 3.7380006313323975,
+      "grad_norm": 3.706770181655884,
       "learning_rate": 1.1834122274818321e-05,
-      "loss": 1.2608,
+      "loss": 1.2626,
       "step": 7900
     },
     {
       "epoch": 0.76,
-      "eval_loss": 1.1713016033172607,
-      "eval_runtime": 27.2356,
-      "eval_samples_per_second": 18.358,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.1713511943817139,
+      "eval_runtime": 27.2512,
+      "eval_samples_per_second": 18.348,
+      "eval_steps_per_second": 2.312,
       "step": 7900
     },
     {
       "epoch": 0.77,
-      "grad_norm": 5.135042190551758,
+      "grad_norm": 4.9828667640686035,
       "learning_rate": 1.1667444496299753e-05,
-      "loss": 1.3132,
+      "loss": 1.3136,
       "step": 8000
     },
     {
       "epoch": 0.77,
-      "eval_loss": 1.1695390939712524,
-      "eval_runtime": 27.2322,
-      "eval_samples_per_second": 18.361,
+      "eval_loss": 1.1698399782180786,
+      "eval_runtime": 27.2353,
+      "eval_samples_per_second": 18.358,
       "eval_steps_per_second": 2.313,
       "step": 8000
     },
     {
       "epoch": 0.78,
-      "grad_norm": 3.053743600845337,
+      "grad_norm": 2.9339020252227783,
       "learning_rate": 1.1500766717781187e-05,
-      "loss": 1.312,
+      "loss": 1.3125,
       "step": 8100
     },
     {
       "epoch": 0.78,
-      "eval_loss": 1.1722323894500732,
-      "eval_runtime": 27.2386,
-      "eval_samples_per_second": 18.356,
+      "eval_loss": 1.171270489692688,
+      "eval_runtime": 27.2365,
+      "eval_samples_per_second": 18.358,
       "eval_steps_per_second": 2.313,
       "step": 8100
     },
     {
       "epoch": 0.79,
-      "grad_norm": 11.018885612487793,
+      "grad_norm": 10.367125511169434,
       "learning_rate": 1.1334088939262619e-05,
-      "loss": 1.3056,
+      "loss": 1.3017,
       "step": 8200
     },
     {
       "epoch": 0.79,
-      "eval_loss": 1.1667100191116333,
-      "eval_runtime": 27.2417,
-      "eval_samples_per_second": 18.354,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.1658000946044922,
+      "eval_runtime": 27.2436,
+      "eval_samples_per_second": 18.353,
+      "eval_steps_per_second": 2.312,
       "step": 8200
     },
     {
       "epoch": 0.8,
-      "grad_norm": 6.420007228851318,
+      "grad_norm": 6.329850196838379,
       "learning_rate": 1.116741116074405e-05,
-      "loss": 1.2606,
+      "loss": 1.2615,
       "step": 8300
     },
     {
       "epoch": 0.8,
-      "eval_loss": 1.167235016822815,
-      "eval_runtime": 27.2335,
-      "eval_samples_per_second": 18.36,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.1665682792663574,
+      "eval_runtime": 27.2448,
+      "eval_samples_per_second": 18.352,
+      "eval_steps_per_second": 2.312,
       "step": 8300
     },
     {
       "epoch": 0.81,
-      "grad_norm": 5.9483842849731445,
+      "grad_norm": 5.759530067443848,
       "learning_rate": 1.1000733382225482e-05,
-      "loss": 1.3797,
+      "loss": 1.3781,
       "step": 8400
     },
     {
       "epoch": 0.81,
-      "eval_loss": 1.1669018268585205,
-      "eval_runtime": 27.2379,
-      "eval_samples_per_second": 18.357,
+      "eval_loss": 1.1660109758377075,
+      "eval_runtime": 27.2383,
+      "eval_samples_per_second": 18.356,
       "eval_steps_per_second": 2.313,
       "step": 8400
     },
     {
       "epoch": 0.82,
-      "grad_norm": 6.5559892654418945,
+      "grad_norm": 6.256554126739502,
       "learning_rate": 1.0834055603706914e-05,
-      "loss": 1.2377,
+      "loss": 1.2429,
       "step": 8500
     },
     {
       "epoch": 0.82,
-      "eval_loss": 1.1625422239303589,
-      "eval_runtime": 27.2397,
-      "eval_samples_per_second": 18.356,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.1605762243270874,
+      "eval_runtime": 27.2435,
+      "eval_samples_per_second": 18.353,
+      "eval_steps_per_second": 2.312,
       "step": 8500
     },
     {
       "epoch": 0.83,
-      "grad_norm": 6.560941219329834,
+      "grad_norm": 6.039596080780029,
       "learning_rate": 1.0667377825188346e-05,
-      "loss": 1.3798,
+      "loss": 1.3783,
       "step": 8600
     },
     {
       "epoch": 0.83,
-      "eval_loss": 1.162017822265625,
-      "eval_runtime": 27.2403,
-      "eval_samples_per_second": 18.355,
-      "eval_steps_per_second": 2.313,
+      "eval_loss": 1.1602510213851929,
+      "eval_runtime": 27.2519,
+      "eval_samples_per_second": 18.347,
+      "eval_steps_per_second": 2.312,
       "step": 8600
     },
     {
       "epoch": 0.84,
-      "grad_norm": 10.945527076721191,
+      "grad_norm": 12.013235092163086,
       "learning_rate": 1.0500700046669778e-05,
-      "loss": 1.2488,
+      "loss": 1.247,
       "step": 8700
     },
     {
       "epoch": 0.84,
-      "eval_loss": 1.15947425365448,
-      "eval_runtime": 27.2348,
-      "eval_samples_per_second": 18.359,
+      "eval_loss": 1.1587607860565186,
+      "eval_runtime": 27.2412,
+      "eval_samples_per_second": 18.355,
       "eval_steps_per_second": 2.313,
       "step": 8700
     },
     {
       "epoch": 0.85,
-      "grad_norm": 3.0299010276794434,
+      "grad_norm": 2.92677640914917,
       "learning_rate": 1.0334022268151212e-05,
-      "loss": 1.2707,
+      "loss": 1.2676,
       "step": 8800
     },
     {
       "epoch": 0.85,
-      "eval_loss": 1.1578799486160278,
-      "eval_runtime": 27.2313,
-      "eval_samples_per_second": 18.361,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.157351016998291,
+      "eval_runtime": 27.2387,
+      "eval_samples_per_second": 18.356,
+      "eval_steps_per_second": 2.313,
       "step": 8800
     },
     {
       "epoch": 0.86,
-      "grad_norm": 2.693074941635132,
+      "grad_norm": 2.9594922065734863,
       "learning_rate": 1.0167344489632642e-05,
-      "loss": 1.163,
+      "loss": 1.1611,
       "step": 8900
     },
     {
       "epoch": 0.86,
-      "eval_loss": 1.156869649887085,
-      "eval_runtime": 27.2425,
-      "eval_samples_per_second": 18.354,
+      "eval_loss": 1.1563640832901,
+      "eval_runtime": 27.2391,
+      "eval_samples_per_second": 18.356,
       "eval_steps_per_second": 2.313,
       "step": 8900
     },
     {
       "epoch": 0.87,
-      "grad_norm": 10.922174453735352,
+      "grad_norm": 11.393871307373047,
       "learning_rate": 1.0000666711114074e-05,
-      "loss": 1.2786,
+      "loss": 1.2814,
       "step": 9000
     },
     {
       "epoch": 0.87,
-      "eval_loss": 1.154771089553833,
-      "eval_runtime": 27.2049,
-      "eval_samples_per_second": 18.379,
-      "eval_steps_per_second": 2.316,
+      "eval_loss": 1.1527520418167114,
+      "eval_runtime": 27.2403,
+      "eval_samples_per_second": 18.355,
+      "eval_steps_per_second": 2.313,
       "step": 9000
     },
     {
       "epoch": 0.88,
-      "grad_norm": 12.848760604858398,
+      "grad_norm": 13.028400421142578,
       "learning_rate": 9.833988932595508e-06,
-      "loss": 1.2738,
+      "loss": 1.2742,
       "step": 9100
     },
     {
       "epoch": 0.88,
-      "eval_loss": 1.1499338150024414,
-      "eval_runtime": 27.2148,
-      "eval_samples_per_second": 18.372,
-      "eval_steps_per_second": 2.315,
+      "eval_loss": 1.147485375404358,
+      "eval_runtime": 27.2339,
+      "eval_samples_per_second": 18.359,
+      "eval_steps_per_second": 2.313,
       "step": 9100
     },
     {
       "epoch": 0.89,
-      "grad_norm": 5.136310577392578,
+      "grad_norm": 5.377606391906738,
       "learning_rate": 9.667311154076938e-06,
-      "loss": 1.3132,
+      "loss": 1.315,
       "step": 9200
     },
     {
       "epoch": 0.89,
-      "eval_loss": 1.1506128311157227,
-      "eval_runtime": 27.2118,
-      "eval_samples_per_second": 18.374,
-      "eval_steps_per_second": 2.315,
+      "eval_loss": 1.1482326984405518,
+      "eval_runtime": 27.255,
+      "eval_samples_per_second": 18.345,
+      "eval_steps_per_second": 2.311,
       "step": 9200
     },
     {
       "epoch": 0.9,
-      "grad_norm": 6.445253372192383,
+      "grad_norm": 6.8872199058532715,
       "learning_rate": 9.500633375558372e-06,
-      "loss": 1.292,
+      "loss": 1.2934,
       "step": 9300
     },
     {
       "epoch": 0.9,
-      "eval_loss": 1.1501805782318115,
-      "eval_runtime": 27.2291,
-      "eval_samples_per_second": 18.363,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.1486388444900513,
+      "eval_runtime": 27.2466,
+      "eval_samples_per_second": 18.351,
+      "eval_steps_per_second": 2.312,
       "step": 9300
     },
     {
       "epoch": 0.91,
-      "grad_norm": 7.888484477996826,
+      "grad_norm": 8.581022262573242,
       "learning_rate": 9.333955597039803e-06,
-      "loss": 1.3091,
+      "loss": 1.3053,
       "step": 9400
     },
     {
       "epoch": 0.91,
-      "eval_loss": 1.1491880416870117,
-      "eval_runtime": 27.1957,
-      "eval_samples_per_second": 18.385,
-      "eval_steps_per_second": 2.317,
+      "eval_loss": 1.1475136280059814,
+      "eval_runtime": 27.2485,
+      "eval_samples_per_second": 18.35,
+      "eval_steps_per_second": 2.312,
       "step": 9400
     },
     {
       "epoch": 0.92,
-      "grad_norm": 6.232168674468994,
+      "grad_norm": 6.223934173583984,
       "learning_rate": 9.167277818521235e-06,
-      "loss": 1.2423,
+      "loss": 1.2461,
       "step": 9500
     },
     {
       "epoch": 0.92,
-      "eval_loss": 1.1463291645050049,
-      "eval_runtime": 27.194,
-      "eval_samples_per_second": 18.386,
-      "eval_steps_per_second": 2.317,
+      "eval_loss": 1.1443854570388794,
+      "eval_runtime": 27.25,
+      "eval_samples_per_second": 18.349,
+      "eval_steps_per_second": 2.312,
       "step": 9500
     },
     {
       "epoch": 0.93,
-      "grad_norm": 8.130847930908203,
+      "grad_norm": 7.77838659286499,
       "learning_rate": 9.000600040002667e-06,
-      "loss": 1.2448,
+      "loss": 1.2461,
       "step": 9600
     },
     {
       "epoch": 0.93,
-      "eval_loss": 1.1413744688034058,
-      "eval_runtime": 27.1976,
-      "eval_samples_per_second": 18.384,
-      "eval_steps_per_second": 2.316,
+      "eval_loss": 1.1398392915725708,
+      "eval_runtime": 27.2344,
+      "eval_samples_per_second": 18.359,
+      "eval_steps_per_second": 2.313,
       "step": 9600
     },
     {
       "epoch": 0.94,
-      "grad_norm": 6.337840557098389,
+      "grad_norm": 6.271127223968506,
       "learning_rate": 8.8339222614841e-06,
-      "loss": 1.2572,
+      "loss": 1.2545,
       "step": 9700
     },
     {
       "epoch": 0.94,
-      "eval_loss": 1.1409308910369873,
-      "eval_runtime": 27.2205,
-      "eval_samples_per_second": 18.369,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.139143466949463,
+      "eval_runtime": 27.2452,
+      "eval_samples_per_second": 18.352,
+      "eval_steps_per_second": 2.312,
       "step": 9700
     },
     {
       "epoch": 0.95,
-      "grad_norm": 5.349727153778076,
+      "grad_norm": 5.507913112640381,
       "learning_rate": 8.667244482965531e-06,
-      "loss": 1.2629,
+      "loss": 1.265,
       "step": 9800
     },
     {
       "epoch": 0.95,
-      "eval_loss": 1.1383512020111084,
-      "eval_runtime": 27.2233,
-      "eval_samples_per_second": 18.367,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.1366428136825562,
+      "eval_runtime": 27.2432,
+      "eval_samples_per_second": 18.353,
+      "eval_steps_per_second": 2.312,
       "step": 9800
     },
     {
       "epoch": 0.96,
-      "grad_norm": 13.291428565979004,
+      "grad_norm": 13.688467025756836,
       "learning_rate": 8.500566704446963e-06,
-      "loss": 1.3713,
+      "loss": 1.3709,
       "step": 9900
     },
     {
       "epoch": 0.96,
-      "eval_loss": 1.1351414918899536,
-      "eval_runtime": 27.226,
-      "eval_samples_per_second": 18.365,
-      "eval_steps_per_second": 2.314,
+      "eval_loss": 1.133713722229004,
+      "eval_runtime": 27.2421,
+      "eval_samples_per_second": 18.354,
+      "eval_steps_per_second": 2.313,
       "step": 9900
-    },
-    {
-      "epoch": 0.97,
-      "grad_norm": 7.407767295837402,
-      "learning_rate": 8.333888925928397e-06,
-      "loss": 1.2711,
-      "step": 10000
-    },
-    {
-      "epoch": 0.97,
-      "eval_loss": 1.134214162826538,
-      "eval_runtime": 27.214,
-      "eval_samples_per_second": 18.373,
-      "eval_steps_per_second": 2.315,
-      "step": 10000
-    },
-    {
-      "epoch": 0.97,
-      "grad_norm": 6.4951348304748535,
-      "learning_rate": 8.167211147409829e-06,
-      "loss": 1.2371,
-      "step": 10100
-    },
-    {
-      "epoch": 0.97,
-      "eval_loss": 1.1317152976989746,
-      "eval_runtime": 27.2127,
-      "eval_samples_per_second": 18.374,
-      "eval_steps_per_second": 2.315,
-      "step": 10100
-    },
-    {
-      "epoch": 0.98,
-      "grad_norm": 5.784173488616943,
-      "learning_rate": 8.000533368891259e-06,
-      "loss": 1.2729,
-      "step": 10200
-    },
-    {
-      "epoch": 0.98,
-      "eval_loss": 1.1299734115600586,
-      "eval_runtime": 27.228,
-      "eval_samples_per_second": 18.363,
-      "eval_steps_per_second": 2.314,
-      "step": 10200
-    },
-    {
-      "epoch": 0.99,
-      "grad_norm": 5.597267150878906,
-      "learning_rate": 7.833855590372693e-06,
-      "loss": 1.2332,
-      "step": 10300
-    },
-    {
-      "epoch": 0.99,
-      "eval_loss": 1.1279923915863037,
-      "eval_runtime": 27.2092,
-      "eval_samples_per_second": 18.376,
-      "eval_steps_per_second": 2.315,
-      "step": 10300
     }
   ],
   "logging_steps": 100,
@@ -1559,7 +1499,7 @@
   "num_input_tokens_seen": 0,
   "num_train_epochs": 2,
   "save_steps": 100,
-  "total_flos": 3.888476623847424e+17,
+  "total_flos": 3.73253167214592e+17,
   "train_batch_size": 2,
   "trial_name": null,
   "trial_params": null

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1e274454bab8db82ffc2eb95b83319df0a257ebd888203fc0f1437e4638ddb59
+oid sha256:edb1c01f29b4b99f667656f692cfd48ca3f425497b977065e9aa1d26c47f8507
 size 4920