chore: update magic parameters for finetune ➿➿➿➿➿➿

finetunes/deepseek/ds_config_zero2.json

@@ -0,0 +1,50 @@
+{
+  "fp16": {
+    "enabled": "auto",
+    "loss_scale": 0,
+    "loss_scale_window": 1000,
+    "initial_scale_power": 16,
+    "hysteresis": 2,
+    "min_loss_scale": 1
+  },
+
+  "optimizer": {
+    "type": "AdamW",
+    "params": {
+      "lr": "auto",
+      "betas": "auto",
+      "eps": "auto",
+      "weight_decay": "auto"
+    }
+  },
+
+  "scheduler": {
+    "type": "WarmupLR",
+    "params": {
+      "warmup_min_lr": "auto",
+      "warmup_max_lr": "auto",
+      "warmup_num_steps": "auto"
+    }
+  },
+
+  "zero_optimization": {
+    "stage": 2,
+    "offload_optimizer": {
+      "device": "cpu",
+      "pin_memory": true
+    },
+    "allgather_partitions": true,
+    "allgather_bucket_size": 2e8,
+    "overlap_comm": true,
+    "reduce_scatter": true,
+    "reduce_bucket_size": 2e8,
+    "contiguous_gradients": true
+  },
+
+  "gradient_accumulation_steps": "auto",
+  "gradient_clipping": "auto",
+  "steps_per_print": 2000,
+  "train_batch_size": "auto",
+  "train_micro_batch_size_per_gpu": "auto",
+  "wall_clock_breakdown": false
+}

finetunes/deepseek/ds_config_zero3.json

@@ -0,0 +1,51 @@
+{
+  "bf16": {
+    "enabled": "auto"
+  },
+  "optimizer": {
+    "type": "AdamW",
+    "params": {
+      "lr": "auto",
+      "betas": "auto",
+      "eps": "auto",
+      "weight_decay": "auto"
+    }
+  },
+
+  "scheduler": {
+    "type": "WarmupLR",
+    "params": {
+      "warmup_min_lr": "auto",
+      "warmup_max_lr": "auto",
+      "warmup_num_steps": "auto"
+    }
+  },
+
+  "zero_optimization": {
+    "stage": 3,
+    "offload_optimizer": {
+      "device": "cpu",
+      "pin_memory": true
+    },
+    "offload_param": {
+      "device": "cpu",
+      "pin_memory": true
+    },
+    "overlap_comm": true,
+    "contiguous_gradients": true,
+    "sub_group_size": 1e9,
+    "reduce_bucket_size": "auto",
+    "stage3_prefetch_bucket_size": "auto",
+    "stage3_param_persistence_threshold": "auto",
+    "stage3_max_live_parameters": 1e9,
+    "stage3_max_reuse_distance": 1e9,
+    "stage3_gather_16bit_weights_on_model_save": true
+  },
+
+  "gradient_accumulation_steps": "auto",
+  "gradient_clipping": "auto",
+  "steps_per_print": 20,
+  "train_batch_size": "auto",
+  "train_micro_batch_size_per_gpu": "auto",
+  "wall_clock_breakdown": false
+}

finetunes/deepseek/finetune.ipynb

@@ -85,22 +85,22 @@
     "    --model_name_or_path $MODEL_PATH \\\n",
     "    --data_path $DATA_PATH \\\n",
     "    --output_dir $OUTPUT_PATH \\\n",
-    "    --num_train_epochs 3 \\\n",
-    "    --model_max_length 1024 \\\n",
-    "    --per_device_train_batch_size 2 \\\n",
+    "    --num_train_epochs 4 \\\n",
+    "    --model_max_length 512 \\\n",
+    "    --per_device_train_batch_size 16 \\\n",
     "    --per_device_eval_batch_size 1 \\\n",
     "    --gradient_accumulation_steps 4 \\\n",
     "    --evaluation_strategy \"no\" \\\n",
     "    --save_strategy \"steps\" \\\n",
-    "    --save_steps 100 \\\n",
+    "    --save_steps 50 \\\n",
     "    --save_total_limit 100 \\\n",
     "    --learning_rate 2e-5 \\\n",
     "    --warmup_steps 10 \\\n",
     "    --logging_steps 1 \\\n",
     "    --lr_scheduler_type \"cosine\" \\\n",
     "    --gradient_checkpointing True \\\n",
     "    --report_to \"tensorboard\" \\\n",
-    "    --deepspeed configs/ds_config_zero3.json \\\n",
+    "    --deepspeed configs/ds_config_zero2.json \\\n",
     "    --bf16 True"
    ]
   }

unit-cli/src/main/kotlin/cc/unitmesh/runner/Picker.kt

@@ -17,7 +17,7 @@ private val logger = org.slf4j.LoggerFactory.getLogger(PickerCommand::class.java
 fun main(args: Array<String>) = PickerCommand().main(args)
 
 class PickerCommand : CliktCommand() {
-    val completionTypeSize by option(help = "Limit each CompletionType size").int().default(1000)
+    val completionTypeSize by option(help = "Limit each CompletionType size").int().default(500)
 
     override fun run() {
         val outputDir = File("datasets" + File.separator + "origin")