初始化项目，由ModelHub XC社区提供模型

Model: burtenshaw/Qwen1.5-0.5B-dpo-mix-7k
Source: Original Platform

checkpoint-313/trainer_state.json

@@ -0,0 +1,71 @@
+{
+  "best_metric": 1.1438744068145752,
+  "best_model_checkpoint": "./checkpoints/dpo-mix-7k/Qwen1.5-0.5B-dpo-mix-7k-lambda1.0-ORPO-29-9-49/checkpoint-313",
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 313,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.16,
+      "grad_norm": 616.0,
+      "learning_rate": 2.5e-06,
+      "loss": 15.2717,
+      "step": 50
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 5.59375,
+      "learning_rate": 5e-06,
+      "loss": 4.0109,
+      "step": 100
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 5.65625,
+      "learning_rate": 4.889350784534168e-06,
+      "loss": 1.2041,
+      "step": 150
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 4.90625,
+      "learning_rate": 4.5671977372432355e-06,
+      "loss": 1.1597,
+      "step": 200
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 43.5,
+      "learning_rate": 4.062057643681335e-06,
+      "loss": 1.1542,
+      "step": 250
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 9.8125,
+      "learning_rate": 3.4186451878908393e-06,
+      "loss": 1.1222,
+      "step": 300
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.1438744068145752,
+      "eval_runtime": 34.1359,
+      "eval_samples_per_second": 16.551,
+      "eval_steps_per_second": 1.055,
+      "step": 313
+    }
+  ],
+  "logging_steps": 50,
+  "max_steps": 626,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "total_flos": 4744684177981440.0,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}