Training in progress, step 40, checkpoint

checkpoint-40/adapter_config.json

@@ -23,11 +23,11 @@
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "k_proj",
     "v_proj",
-    "o_proj",
-    "gate_proj",
+    "k_proj",
     "q_proj",
+    "gate_proj",
+    "o_proj",
     "down_proj",
     "up_proj"
   ],

checkpoint-40/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:541f8bcd61c4fb0afe2a99c6edd1e3a00c98ac168d31f1992423a006e64d1991
+oid sha256:5911c4749a01fd9ddfe397b9b568b17639e808ca02e971c065c141c9ece7a8ea
 size 145287696

checkpoint-40/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e8c42b3f6ba7f0de4ecffe94b0c05ccbe531c7b8a54bc843c1502d335fa645cc
+oid sha256:4ad5617c821f3e508ff5d34f6555e0d8a59bdd36666e6fadc17712c59bb4473f
 size 290833618

checkpoint-40/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9e93e3653932bd1c6d7fb4effd7301829398be43b5479a9a2f9e3be2bde051e1
+oid sha256:0f29b5271137232849fdde493db9fbf7a5966f71bdd58624a375220dcdcc1ecd
 size 1064

checkpoint-40/trainer_state.json

@@ -10,69 +10,69 @@
   "log_history": [
     {
       "epoch": 0.021739130434782608,
-      "grad_norm": 2.138256549835205,
-      "learning_rate": 4.5e-05,
-      "loss": 2.3657,
+      "grad_norm": 2.2049312591552734,
+      "learning_rate": 4.963768115942029e-05,
+      "loss": 2.3459,
       "step": 10
     },
     {
       "epoch": 0.021739130434782608,
-      "eval_loss": 1.5849242210388184,
-      "eval_runtime": 9.6002,
-      "eval_samples_per_second": 49.999,
-      "eval_steps_per_second": 1.562,
+      "eval_loss": 1.5330805778503418,
+      "eval_runtime": 9.6608,
+      "eval_samples_per_second": 49.685,
+      "eval_steps_per_second": 1.553,
       "step": 10
     },
     {
       "epoch": 0.043478260869565216,
-      "grad_norm": 1.1203280687332153,
-      "learning_rate": 4e-05,
-      "loss": 1.2125,
+      "grad_norm": 1.1927831172943115,
+      "learning_rate": 4.9275362318840584e-05,
+      "loss": 1.1669,
       "step": 20
     },
     {
       "epoch": 0.043478260869565216,
-      "eval_loss": 1.04584538936615,
-      "eval_runtime": 9.677,
-      "eval_samples_per_second": 49.602,
-      "eval_steps_per_second": 1.55,
+      "eval_loss": 1.0271039009094238,
+      "eval_runtime": 9.7215,
+      "eval_samples_per_second": 49.375,
+      "eval_steps_per_second": 1.543,
       "step": 20
     },
     {
       "epoch": 0.06521739130434782,
-      "grad_norm": 1.064527988433838,
-      "learning_rate": 3.5e-05,
-      "loss": 0.867,
+      "grad_norm": 1.0509072542190552,
+      "learning_rate": 4.891304347826087e-05,
+      "loss": 0.8327,
       "step": 30
     },
     {
       "epoch": 0.06521739130434782,
-      "eval_loss": 0.9389117360115051,
-      "eval_runtime": 9.7352,
-      "eval_samples_per_second": 49.305,
-      "eval_steps_per_second": 1.541,
+      "eval_loss": 0.9039635062217712,
+      "eval_runtime": 9.7818,
+      "eval_samples_per_second": 49.071,
+      "eval_steps_per_second": 1.533,
       "step": 30
     },
     {
       "epoch": 0.08695652173913043,
-      "grad_norm": 1.1056835651397705,
-      "learning_rate": 3e-05,
-      "loss": 0.7879,
+      "grad_norm": 0.8860757350921631,
+      "learning_rate": 4.855072463768116e-05,
+      "loss": 0.731,
       "step": 40
     },
     {
       "epoch": 0.08695652173913043,
-      "eval_loss": 0.8724283576011658,
-      "eval_runtime": 9.8116,
-      "eval_samples_per_second": 48.922,
+      "eval_loss": 0.8067704439163208,
+      "eval_runtime": 9.8118,
+      "eval_samples_per_second": 48.921,
       "eval_steps_per_second": 1.529,
       "step": 40
     }
   ],
   "logging_steps": 10,
-  "max_steps": 100,
+  "max_steps": 1380,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 1,
+  "num_train_epochs": 3,
   "save_steps": 10,
   "stateful_callbacks": {
     "TrainerControl": {

checkpoint-40/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:17a1a7c75c85fbba54f1e97b2cee73c167834890dd0df62950989daaf8975814
+oid sha256:1cdf2cbd96a7d8012e1f3da0569783099ba26a3ecbec680ff36cde09793d0889
 size 5816