Training in progress, step 200, checkpoint

last-checkpoint/adapter_config.json

@@ -13,7 +13,7 @@
   "layers_pattern": null,
   "layers_to_transform": null,
   "loftq_config": {},
-  "lora_alpha": 12,
+  "lora_alpha": 24,
   "lora_bias": false,
   "lora_dropout": 0.25,
   "megatron_config": null,
@@ -21,17 +21,17 @@
   "modules_to_save": null,
   "peft_type": "LORA",
   "qalora_group_size": 16,
-  "r": 8,
+  "r": 12,
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
-    "down_proj",
-    "k_proj",
+    "q_proj",
     "up_proj",
-    "gate_proj",
+    "k_proj",
+    "o_proj",
     "v_proj",
-    "q_proj",
-    "o_proj"
+    "gate_proj",
+    "down_proj"
   ],
   "target_parameters": null,
   "task_type": "SEQ_2_SEQ_LM",

last-checkpoint/adapter_model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a3b6cec6bb381d06a610189bd79a8deb709e3c5d2fdf5d819a60d28011decc79
-size 5544448
+oid sha256:cb7edc022a38beb145ffb51c73bf9826a63972fa7eb7866f4c9659074c1a8c43
+size 8297288

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9da034081b14a2f97a258a7f15ae29da09a4f53994ae505db3cd996ec1942269
-size 11256139
+oid sha256:3ff0ff603551669ef970f84399569e9870a0c8597521fcef27f66df3c6f5ef22
+size 16761163

last-checkpoint/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:97026289a5ff3e49253420f06290674139f61467712a2029e3c1001c12a71fb4
+oid sha256:fc90d0ce81a109e8e9dd4fc59d9ed7b59e19eb969b6e5c7939731348099ada0d
 size 14645

last-checkpoint/scaler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:43f4f16c27b661199b56bd3230b24cfb3ac0ec6e2d7124e3005fac49dfb22892
+oid sha256:cfc079ff01029f23f0ea697b8a7da7f71ea1510cb9be263645b8bef9dfeb3f70
 size 1383

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f00d7d2ea6eff43e89aa523cc15b09467f552729fa90e936a173b939c5db07ba
+oid sha256:66df8c272ec6ea7fc32240052101ceb1261559eaab8e63c8d4d99dac2464379a
 size 1465

last-checkpoint/trainer_state.json

@@ -1,179 +1,64 @@
 {
   "best_global_step": 200,
-  "best_metric": 4.437816143035889,
+  "best_metric": 0.6018396615982056,
   "best_model_checkpoint": "./t5gemma-math-corrector/checkpoint-200",
-  "epoch": 3.0,
+  "epoch": 0.6415396952686447,
   "eval_steps": 200,
-  "global_step": 888,
+  "global_step": 200,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.1693480101608806,
-      "grad_norm": 1.3522921800613403,
-      "learning_rate": 5.50561797752809e-05,
-      "loss": 2.2844,
+      "epoch": 0.16038492381716118,
+      "grad_norm": 2.080294370651245,
+      "learning_rate": 5.212765957446809e-05,
+      "loss": 1.8373,
       "step": 50
     },
     {
-      "epoch": 0.3386960203217612,
-      "grad_norm": 0.5189423561096191,
-      "learning_rate": 9.874843554443054e-05,
-      "loss": 0.0374,
+      "epoch": 0.32076984763432237,
+      "grad_norm": 1.3640859127044678,
+      "learning_rate": 9.94061757719715e-05,
+      "loss": 0.028,
       "step": 100
     },
     {
-      "epoch": 0.5080440304826418,
-      "grad_norm": 0.2796284258365631,
-      "learning_rate": 9.249061326658323e-05,
-      "loss": 0.0167,
+      "epoch": 0.48115477145148355,
+      "grad_norm": 0.47297751903533936,
+      "learning_rate": 9.346793349168646e-05,
+      "loss": 0.0138,
       "step": 150
     },
     {
-      "epoch": 0.6773920406435224,
-      "grad_norm": 0.5644184350967407,
-      "learning_rate": 8.623279098873592e-05,
-      "loss": 0.0108,
+      "epoch": 0.6415396952686447,
+      "grad_norm": 0.6050183176994324,
+      "learning_rate": 8.752969121140144e-05,
+      "loss": 0.0085,
       "step": 200
     },
     {
-      "epoch": 0.6773920406435224,
-      "eval_loss": 4.437816143035889,
-      "eval_runtime": 10.3595,
-      "eval_samples_per_second": 50.582,
-      "eval_steps_per_second": 12.645,
+      "epoch": 0.6415396952686447,
+      "eval_loss": 0.6018396615982056,
+      "eval_runtime": 5.1929,
+      "eval_samples_per_second": 50.454,
+      "eval_steps_per_second": 12.71,
       "step": 200
-    },
-    {
-      "epoch": 0.8467400508044031,
-      "grad_norm": 0.395366907119751,
-      "learning_rate": 7.997496871088861e-05,
-      "loss": 0.0073,
-      "step": 250
-    },
-    {
-      "epoch": 1.0135478408128704,
-      "grad_norm": 0.14810292422771454,
-      "learning_rate": 7.371714643304131e-05,
-      "loss": 0.0079,
-      "step": 300
-    },
-    {
-      "epoch": 1.182895850973751,
-      "grad_norm": 0.2394929677248001,
-      "learning_rate": 6.7459324155194e-05,
-      "loss": 0.0039,
-      "step": 350
-    },
-    {
-      "epoch": 1.3522438611346317,
-      "grad_norm": 0.4444918930530548,
-      "learning_rate": 6.120150187734669e-05,
-      "loss": 0.0034,
-      "step": 400
-    },
-    {
-      "epoch": 1.3522438611346317,
-      "eval_loss": 7.077005863189697,
-      "eval_runtime": 10.3846,
-      "eval_samples_per_second": 50.459,
-      "eval_steps_per_second": 12.615,
-      "step": 400
-    },
-    {
-      "epoch": 1.5215918712955123,
-      "grad_norm": 0.19818231463432312,
-      "learning_rate": 5.4943679599499376e-05,
-      "loss": 0.0026,
-      "step": 450
-    },
-    {
-      "epoch": 1.690939881456393,
-      "grad_norm": 0.21913927793502808,
-      "learning_rate": 4.8685857321652064e-05,
-      "loss": 0.0027,
-      "step": 500
-    },
-    {
-      "epoch": 1.8602878916172734,
-      "grad_norm": 0.10620615631341934,
-      "learning_rate": 4.242803504380476e-05,
-      "loss": 0.0025,
-      "step": 550
-    },
-    {
-      "epoch": 2.027095681625741,
-      "grad_norm": 0.03891622647643089,
-      "learning_rate": 3.617021276595745e-05,
-      "loss": 0.002,
-      "step": 600
-    },
-    {
-      "epoch": 2.027095681625741,
-      "eval_loss": 7.624501705169678,
-      "eval_runtime": 10.3997,
-      "eval_samples_per_second": 50.386,
-      "eval_steps_per_second": 12.597,
-      "step": 600
-    },
-    {
-      "epoch": 2.1964436917866217,
-      "grad_norm": 0.05978698655962944,
-      "learning_rate": 2.9912390488110137e-05,
-      "loss": 0.0012,
-      "step": 650
-    },
-    {
-      "epoch": 2.365791701947502,
-      "grad_norm": 0.04675230383872986,
-      "learning_rate": 2.365456821026283e-05,
-      "loss": 0.0011,
-      "step": 700
-    },
-    {
-      "epoch": 2.535139712108383,
-      "grad_norm": 0.056206703186035156,
-      "learning_rate": 1.739674593241552e-05,
-      "loss": 0.001,
-      "step": 750
-    },
-    {
-      "epoch": 2.7044877222692634,
-      "grad_norm": 0.11963123828172684,
-      "learning_rate": 1.113892365456821e-05,
-      "loss": 0.0011,
-      "step": 800
-    },
-    {
-      "epoch": 2.7044877222692634,
-      "eval_loss": 7.175710201263428,
-      "eval_runtime": 10.392,
-      "eval_samples_per_second": 50.423,
-      "eval_steps_per_second": 12.606,
-      "step": 800
-    },
-    {
-      "epoch": 2.873835732430144,
-      "grad_norm": 0.19036388397216797,
-      "learning_rate": 4.881101376720902e-06,
-      "loss": 0.0012,
-      "step": 850
     }
   ],
   "logging_steps": 50,
-  "max_steps": 888,
+  "max_steps": 936,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 3,
   "save_steps": 200,
   "stateful_callbacks": {
     "EarlyStoppingCallback": {
       "args": {
-        "early_stopping_patience": 5,
+        "early_stopping_patience": 3,
         "early_stopping_threshold": 0.0
       },
       "attributes": {
-        "early_stopping_patience_counter": 3
+        "early_stopping_patience_counter": 0
       }
     },
     "TrainerControl": {
@@ -182,12 +67,12 @@
         "should_evaluate": false,
         "should_log": false,
         "should_save": true,
-        "should_training_stop": true
+        "should_training_stop": false
       },
       "attributes": {}
     }
   },
-  "total_flos": 3379504493887488.0,
+  "total_flos": 773230008729600.0,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null

last-checkpoint/training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8c67ac096c206b245d26f775afa3fb8ffa810038da5c0f227d1b18811d5d602b
+oid sha256:a5e19188bdca9ed7ec3f5485c2058b4803117cec11f54ee11e729dcda3bd5b01
 size 5905