初始化项目，由ModelHub XC社区提供模型

Model: PKU-Alignment/ProgressGym-HistLlama3-8B-C021-instruct-v0.2
Source: Original Platform

trainer_state.json

@@ -0,0 +1,65 @@
+{
+  "best_metric": 0.8208221793174744,
+  "best_model_checkpoint": "./output/training_results/C021_random_sample_Meta-Llama-3-8B_instruct_20240729_032429/checkpoint-10",
+  "epoch": 0.20833333333333334,
+  "eval_steps": 5,
+  "global_step": 10,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.020833333333333332,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.9812,
+      "step": 1
+    },
+    {
+      "epoch": 0.0625,
+      "grad_norm": 0.0,
+      "learning_rate": 0.0,
+      "loss": 0.9339,
+      "step": 3
+    },
+    {
+      "epoch": 0.10416666666666667,
+      "eval_loss": 1.004961609840393,
+      "eval_runtime": 0.8973,
+      "eval_samples_per_second": 378.933,
+      "eval_steps_per_second": 3.344,
+      "step": 5
+    },
+    {
+      "epoch": 0.125,
+      "grad_norm": 12.659072715759642,
+      "learning_rate": 2.25e-06,
+      "loss": 1.0561,
+      "step": 6
+    },
+    {
+      "epoch": 0.1875,
+      "grad_norm": 5.826176159289086,
+      "learning_rate": 2.865909090909091e-06,
+      "loss": 0.8863,
+      "step": 9
+    },
+    {
+      "epoch": 0.20833333333333334,
+      "eval_loss": 0.8208221793174744,
+      "eval_runtime": 0.8762,
+      "eval_samples_per_second": 388.052,
+      "eval_steps_per_second": 3.424,
+      "step": 10
+    }
+  ],
+  "logging_steps": 3,
+  "max_steps": 48,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 10,
+  "total_flos": 183002726400.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}