Training in progress, epoch 1

all_results.json

@@ -0,0 +1,18 @@
+{
+    "epoch": 9.0,
+    "eval_gen_len": 294.2,
+    "eval_loss": 2.3246188163757324,
+    "eval_rouge1": 24.0821,
+    "eval_rouge2": 4.2604,
+    "eval_rougeL": 12.4489,
+    "eval_rougeLsum": 19.152,
+    "eval_runtime": 24.5774,
+    "eval_samples": 10,
+    "eval_samples_per_second": 0.407,
+    "eval_steps_per_second": 0.407,
+    "train_loss": 2.465799797905816,
+    "train_runtime": 251.0487,
+    "train_samples": 10,
+    "train_samples_per_second": 0.398,
+    "train_steps_per_second": 0.398
+}

eval_results.json

@@ -0,0 +1,13 @@
+{
+    "epoch": 9.0,
+    "eval_gen_len": 294.2,
+    "eval_loss": 2.3246188163757324,
+    "eval_rouge1": 24.0821,
+    "eval_rouge2": 4.2604,
+    "eval_rougeL": 12.4489,
+    "eval_rougeLsum": 19.152,
+    "eval_runtime": 24.5774,
+    "eval_samples": 10,
+    "eval_samples_per_second": 0.407,
+    "eval_steps_per_second": 0.407
+}

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:22905e0021122dcb02d60edd9d9d2456b7372972825d1d29936cf966e9324ed0
+oid sha256:d60f1cf73003e01c263cd0f1467478ae6a7404aeaa6ea121af69f87d2748b025
 size 501748312

runs/Feb21_17-01-57_kogecha/events.out.tfevents.1708502860.kogecha

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:110a20ee9c70586920947f209717642ae401cc8e8897af7a65f6979f7db3d078
+size 555

runs/Feb21_17-15-41_kogecha/events.out.tfevents.1708503352.kogecha

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a8fdbd323934205d13ae6c34b400cf35b2c84985156ab7da924d1081e47a02b8
+size 6150

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 9.0,
+    "train_loss": 2.465799797905816,
+    "train_runtime": 251.0487,
+    "train_samples": 10,
+    "train_samples_per_second": 0.398,
+    "train_steps_per_second": 0.398
+}

trainer_state.json

@@ -0,0 +1,147 @@
+{
+  "best_metric": 24.0821,
+  "best_model_checkpoint": "saved/tobyoki-pairwise/bart-base-japanese/BaseModel/checkpoint-60",
+  "epoch": 9.0,
+  "eval_steps": 500,
+  "global_step": 90,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 1.0,
+      "eval_gen_len": 300.0,
+      "eval_loss": 2.7064690589904785,
+      "eval_rouge1": 17.4135,
+      "eval_rouge2": 1.8293,
+      "eval_rougeL": 10.3284,
+      "eval_rougeLsum": 13.7135,
+      "eval_runtime": 24.7791,
+      "eval_samples_per_second": 0.404,
+      "eval_steps_per_second": 0.404,
+      "step": 10
+    },
+    {
+      "epoch": 2.0,
+      "eval_gen_len": 300.0,
+      "eval_loss": 2.4982924461364746,
+      "eval_rouge1": 15.5531,
+      "eval_rouge2": 1.4496,
+      "eval_rougeL": 9.0138,
+      "eval_rougeLsum": 12.8628,
+      "eval_runtime": 24.9094,
+      "eval_samples_per_second": 0.401,
+      "eval_steps_per_second": 0.401,
+      "step": 20
+    },
+    {
+      "epoch": 3.0,
+      "eval_gen_len": 300.0,
+      "eval_loss": 2.418119192123413,
+      "eval_rouge1": 17.7832,
+      "eval_rouge2": 2.7959,
+      "eval_rougeL": 9.969,
+      "eval_rougeLsum": 13.5642,
+      "eval_runtime": 25.6031,
+      "eval_samples_per_second": 0.391,
+      "eval_steps_per_second": 0.391,
+      "step": 30
+    },
+    {
+      "epoch": 4.0,
+      "eval_gen_len": 300.0,
+      "eval_loss": 2.3930418491363525,
+      "eval_rouge1": 20.0103,
+      "eval_rouge2": 2.9976,
+      "eval_rougeL": 11.0143,
+      "eval_rougeLsum": 15.4415,
+      "eval_runtime": 24.7517,
+      "eval_samples_per_second": 0.404,
+      "eval_steps_per_second": 0.404,
+      "step": 40
+    },
+    {
+      "epoch": 5.0,
+      "eval_gen_len": 300.0,
+      "eval_loss": 2.332805871963501,
+      "eval_rouge1": 20.5428,
+      "eval_rouge2": 3.6354,
+      "eval_rougeL": 11.3166,
+      "eval_rougeLsum": 16.0252,
+      "eval_runtime": 25.4485,
+      "eval_samples_per_second": 0.393,
+      "eval_steps_per_second": 0.393,
+      "step": 50
+    },
+    {
+      "epoch": 6.0,
+      "eval_gen_len": 294.2,
+      "eval_loss": 2.3246188163757324,
+      "eval_rouge1": 24.0821,
+      "eval_rouge2": 4.2604,
+      "eval_rougeL": 12.4489,
+      "eval_rougeLsum": 19.152,
+      "eval_runtime": 25.6727,
+      "eval_samples_per_second": 0.39,
+      "eval_steps_per_second": 0.39,
+      "step": 60
+    },
+    {
+      "epoch": 7.0,
+      "eval_gen_len": 300.0,
+      "eval_loss": 2.3076653480529785,
+      "eval_rouge1": 20.4387,
+      "eval_rouge2": 3.8043,
+      "eval_rougeL": 11.4571,
+      "eval_rougeLsum": 16.7923,
+      "eval_runtime": 24.5822,
+      "eval_samples_per_second": 0.407,
+      "eval_steps_per_second": 0.407,
+      "step": 70
+    },
+    {
+      "epoch": 8.0,
+      "eval_gen_len": 300.0,
+      "eval_loss": 2.305992603302002,
+      "eval_rouge1": 22.0434,
+      "eval_rouge2": 4.3648,
+      "eval_rougeL": 12.386,
+      "eval_rougeLsum": 18.3828,
+      "eval_runtime": 25.0442,
+      "eval_samples_per_second": 0.399,
+      "eval_steps_per_second": 0.399,
+      "step": 80
+    },
+    {
+      "epoch": 9.0,
+      "eval_gen_len": 289.1,
+      "eval_loss": 2.302980661392212,
+      "eval_rouge1": 20.3362,
+      "eval_rouge2": 3.6623,
+      "eval_rougeL": 10.9234,
+      "eval_rougeLsum": 16.1471,
+      "eval_runtime": 25.3281,
+      "eval_samples_per_second": 0.395,
+      "eval_steps_per_second": 0.395,
+      "step": 90
+    },
+    {
+      "epoch": 9.0,
+      "step": 90,
+      "total_flos": 18151534964736.0,
+      "train_loss": 2.465799797905816,
+      "train_runtime": 251.0487,
+      "train_samples_per_second": 0.398,
+      "train_steps_per_second": 0.398
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 100,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "total_flos": 18151534964736.0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:b75524dc19d9ed3a2c285703e96f7e376466b769483d7cc40215a41a987a632a
+oid sha256:747bc2d34c5bba4bb557a74792defb10cd54f7eb48b5bb5da8c4399f95e2d54a
 size 4731