Training in progress, step 200, checkpoint

last-checkpoint/adapter_config.json

@@ -13,25 +13,25 @@
   "layers_pattern": null,
   "layers_to_transform": null,
   "loftq_config": {},
-  "lora_alpha": 24,
+  "lora_alpha": 12,
   "lora_bias": false,
-  "lora_dropout": 0.2,
+  "lora_dropout": 0.25,
   "megatron_config": null,
   "megatron_core": "megatron.core",
   "modules_to_save": null,
   "peft_type": "LORA",
   "qalora_group_size": 16,
-  "r": 12,
+  "r": 8,
   "rank_pattern": {},
   "revision": null,
   "target_modules": [
+    "down_proj",
     "k_proj",
+    "up_proj",
     "gate_proj",
     "v_proj",
     "q_proj",
-    "up_proj",
-    "o_proj",
-    "down_proj"
+    "o_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 sha256:617b2d52e25dd054d8ff7bc44937c8020709692eb0658c3ae6d664c24b6ccd63
-size 8297288
+oid sha256:fe58d3c03518659248e48a169ce5f7ad1fd9fcba3b7ff3fbc00f64781e12c93e
+size 5544448

last-checkpoint/optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dc4ffe06d55ff62afc27995c7351fa48a7249b047a862dcddc1aaa5b828aa191
-size 16761163
+oid sha256:f2316dc94bc7196071a1ce63417540cdbedc1a0a62a9b75519dc11ce86d36998
+size 11256139

last-checkpoint/rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:26d1df2276f727e3cca661e1a84a63428c1b4328b6b80b1d5325b41c0298da4d
+oid sha256:fc90d0ce81a109e8e9dd4fc59d9ed7b59e19eb969b6e5c7939731348099ada0d
 size 14645

last-checkpoint/scaler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4b89e7f9c8e503d4cfdc04fb61835e00a23616af74eceff894d548f5f90cea56
+oid sha256:f43a86ff0d782e83d298516157d9a03af28a39748fca022b78af2a5421c34200
 size 1383

last-checkpoint/scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b1cfc743553d4d9ba9acea35cf048c15e0c667b2f4210c4a170ee0849112d83b
+oid sha256:e585b45f6f6e75a7af202c134b7e015219af01e7a03b30286dd1941b3e37836f
 size 1465

last-checkpoint/trainer_state.json

@@ -1,172 +1,64 @@
 {
   "best_global_step": 200,
-  "best_metric": 0.15597519278526306,
+  "best_metric": 4.437816143035889,
   "best_model_checkpoint": "./t5gemma-math-corrector/checkpoint-200",
-  "epoch": 2.5645549318364074,
+  "epoch": 0.6773920406435224,
   "eval_steps": 200,
-  "global_step": 800,
+  "global_step": 200,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.16038492381716118,
-      "grad_norm": 1.8715651035308838,
-      "learning_rate": 5.212765957446809e-05,
-      "loss": 1.8361,
+      "epoch": 0.1693480101608806,
+      "grad_norm": 1.3522921800613403,
+      "learning_rate": 5.50561797752809e-05,
+      "loss": 2.2844,
       "step": 50
     },
     {
-      "epoch": 0.32076984763432237,
-      "grad_norm": 2.1524851322174072,
-      "learning_rate": 9.94061757719715e-05,
-      "loss": 0.0277,
+      "epoch": 0.3386960203217612,
+      "grad_norm": 0.5189423561096191,
+      "learning_rate": 9.874843554443054e-05,
+      "loss": 0.0374,
       "step": 100
     },
     {
-      "epoch": 0.48115477145148355,
-      "grad_norm": 0.4908048212528229,
-      "learning_rate": 9.346793349168646e-05,
-      "loss": 0.013,
+      "epoch": 0.5080440304826418,
+      "grad_norm": 0.2796284258365631,
+      "learning_rate": 9.249061326658323e-05,
+      "loss": 0.0167,
       "step": 150
     },
     {
-      "epoch": 0.6415396952686447,
-      "grad_norm": 0.43149054050445557,
-      "learning_rate": 8.752969121140144e-05,
-      "loss": 0.0083,
+      "epoch": 0.6773920406435224,
+      "grad_norm": 0.5644184350967407,
+      "learning_rate": 8.623279098873592e-05,
+      "loss": 0.0108,
       "step": 200
     },
     {
-      "epoch": 0.6415396952686447,
-      "eval_loss": 0.15597519278526306,
-      "eval_runtime": 5.2148,
-      "eval_samples_per_second": 50.242,
-      "eval_steps_per_second": 12.656,
+      "epoch": 0.6773920406435224,
+      "eval_loss": 4.437816143035889,
+      "eval_runtime": 10.3595,
+      "eval_samples_per_second": 50.582,
+      "eval_steps_per_second": 12.645,
       "step": 200
-    },
-    {
-      "epoch": 0.8019246190858059,
-      "grad_norm": 0.18161173164844513,
-      "learning_rate": 8.15914489311164e-05,
-      "loss": 0.0058,
-      "step": 250
-    },
-    {
-      "epoch": 0.9623095429029671,
-      "grad_norm": 0.7308599352836609,
-      "learning_rate": 7.565320665083135e-05,
-      "loss": 0.0044,
-      "step": 300
-    },
-    {
-      "epoch": 1.1218925421010426,
-      "grad_norm": 0.13639520108699799,
-      "learning_rate": 6.971496437054633e-05,
-      "loss": 0.0023,
-      "step": 350
-    },
-    {
-      "epoch": 1.2822774659182037,
-      "grad_norm": 0.07872019708156586,
-      "learning_rate": 6.377672209026129e-05,
-      "loss": 0.0027,
-      "step": 400
-    },
-    {
-      "epoch": 1.2822774659182037,
-      "eval_loss": 0.8356682658195496,
-      "eval_runtime": 5.2047,
-      "eval_samples_per_second": 50.339,
-      "eval_steps_per_second": 12.681,
-      "step": 400
-    },
-    {
-      "epoch": 1.4426623897353648,
-      "grad_norm": 0.3794359266757965,
-      "learning_rate": 5.783847980997625e-05,
-      "loss": 0.0012,
-      "step": 450
-    },
-    {
-      "epoch": 1.603047313552526,
-      "grad_norm": 0.23721610009670258,
-      "learning_rate": 5.190023752969121e-05,
-      "loss": 0.0014,
-      "step": 500
-    },
-    {
-      "epoch": 1.7634322373696873,
-      "grad_norm": 0.12510955333709717,
-      "learning_rate": 4.596199524940617e-05,
-      "loss": 0.0007,
-      "step": 550
-    },
-    {
-      "epoch": 1.9238171611868484,
-      "grad_norm": 0.03999270871281624,
-      "learning_rate": 4.002375296912114e-05,
-      "loss": 0.0009,
-      "step": 600
-    },
-    {
-      "epoch": 1.9238171611868484,
-      "eval_loss": 1.314481496810913,
-      "eval_runtime": 5.2047,
-      "eval_samples_per_second": 50.339,
-      "eval_steps_per_second": 12.681,
-      "step": 600
-    },
-    {
-      "epoch": 2.0834001603849237,
-      "grad_norm": 0.1903328150510788,
-      "learning_rate": 3.408551068883611e-05,
-      "loss": 0.0006,
-      "step": 650
-    },
-    {
-      "epoch": 2.2437850842020852,
-      "grad_norm": 0.0275680311024189,
-      "learning_rate": 2.8147268408551068e-05,
-      "loss": 0.0004,
-      "step": 700
-    },
-    {
-      "epoch": 2.4041700080192463,
-      "grad_norm": 0.07852037996053696,
-      "learning_rate": 2.2209026128266035e-05,
-      "loss": 0.0004,
-      "step": 750
-    },
-    {
-      "epoch": 2.5645549318364074,
-      "grad_norm": 0.08535940945148468,
-      "learning_rate": 1.6270783847980998e-05,
-      "loss": 0.0004,
-      "step": 800
-    },
-    {
-      "epoch": 2.5645549318364074,
-      "eval_loss": 0.740967333316803,
-      "eval_runtime": 5.2046,
-      "eval_samples_per_second": 50.34,
-      "eval_steps_per_second": 12.681,
-      "step": 800
     }
   ],
   "logging_steps": 50,
-  "max_steps": 936,
+  "max_steps": 888,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 3,
   "save_steps": 200,
   "stateful_callbacks": {
     "EarlyStoppingCallback": {
       "args": {
-        "early_stopping_patience": 3,
+        "early_stopping_patience": 5,
         "early_stopping_threshold": 0.0
       },
       "attributes": {
-        "early_stopping_patience_counter": 3
+        "early_stopping_patience_counter": 0
       }
     },
     "TrainerControl": {
@@ -175,12 +67,12 @@
         "should_evaluate": false,
         "should_log": false,
         "should_save": true,
-        "should_training_stop": true
+        "should_training_stop": false
       },
       "attributes": {}
     }
   },
-  "total_flos": 3090020422385664.0,
+  "total_flos": 763083148492800.0,
   "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 sha256:481d9ca92757313e849b38438646d9a43638d96e56b561bb9a27f8af92718d80
+oid sha256:8c67ac096c206b245d26f775afa3fb8ffa810038da5c0f227d1b18811d5d602b
 size 5905