7b

llama-factory/config/models/Qwen2.5-72B-Instruct.yaml

@@ -0,0 +1,43 @@
+### model
+model_name_or_path: Qwen/Qwen2.5-72B-Instruct
+quantization_bit: 4
+
+### method
+stage: sft
+do_train: true
+finetuning_type: lora
+lora_target: all
+
+### dataset
+dataset: alpaca_mgtv_p2
+template: qwen
+cutoff_len: 8192
+max_samples: 25000
+overwrite_cache: true
+preprocessing_num_workers: 16
+
+### output
+output_dir: saves/Qwen2.5-72B-Instruct
+logging_steps: 5
+save_steps: 35
+plot_loss: true
+# overwrite_output_dir: true
+
+### train
+per_device_train_batch_size: 16
+gradient_accumulation_steps: 8
+learning_rate: 1.0e-4
+num_train_epochs: 2.0
+lr_scheduler_type: cosine
+warmup_ratio: 0.1
+bf16: true
+ddp_timeout: 180000000
+
+### eval
+val_size: 0.1
+per_device_eval_batch_size: 1
+eval_strategy: steps
+eval_steps: 35
+
+report_to: wandb
+run_name: Qwen2.5-72B-Instruct_lora_sft

logs/Qwen2.5-7B-Instruct_epoch_1.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3eb21a1e6bba7f6b3162293b01cdf4f80e81044342a1afc90cef71d87cca1d04
+size 17547

logs/Qwen2.5-7B-Instruct_epoch_10.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c18b323927342384b5c2916642be7100cd68558b01afddc576a89f26b5574a38
+size 17397

logs/Qwen2.5-7B-Instruct_epoch_2.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d679717eeb115236b3dbd71be9c0889d7089433d81007032d2c8bc03759fcdc8
+size 17597

logs/Qwen2.5-7B-Instruct_epoch_3.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8190376a1eed3d4cc0fa2304daf6b5905ca4875a0c989085113de79c49a50ec
+size 17490

logs/Qwen2.5-7B-Instruct_epoch_4.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:28c5cfd7853d62cacd618ae27fbb9acb453ac02af0eef1b5d618abb8bdbc6cac
+size 17473

logs/Qwen2.5-7B-Instruct_epoch_5.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e79deb7310a05eb990f0ab88ef28dbe5fbbaab6de5932f95734266889f750cf6
+size 17122

logs/Qwen2.5-7B-Instruct_epoch_6.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbb92e8dc536224b47d905ffc52851bbfcab8e73702461a51a46f8a985158129
+size 17600

logs/Qwen2.5-7B-Instruct_epoch_7.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34aa91ae94e34308c1bcc116cf77da85f053df27f5660cca32aa0f9c2869b9ce
+size 17537

logs/Qwen2.5-7B-Instruct_epoch_8.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7a61e5ccb7a9ae5f2b8e34680783e080f67a11062bcb50f1c6fc99efd226694f
+size 17416

logs/Qwen2.5-7B-Instruct_epoch_9.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c79c85037b52efc95d14b735cb70bae41350275c74da60656d21b661bb01411
+size 17418

scripts/eval-mgtv-qwen2.5.sh

@@ -13,13 +13,13 @@ cat /etc/os-release
 lscpu
 grep MemTotal /proc/meminfo
 
+$BASEDIR/scripts/eval-epochs.sh Qwen Qwen2.5-7B-Instruct
 $BASEDIR/scripts/eval-shots.sh Qwen Qwen2.5-7B-Instruct
 
-$BASEDIR/scripts/eval-epochs.sh Qwen Qwen2.5-7B-Instruct
 
 # $BASEDIR/scripts/eval-shots.sh Qwen Qwen2.5-3B-Instruct
 
-$BASEDIR/scripts/eval-epochs.sh Qwen Qwen2.5-3B-Instruct
+#$BASEDIR/scripts/eval-epochs.sh Qwen Qwen2.5-3B-Instruct
 
 # $BASEDIR/scripts/eval-epochs.sh Qwen Qwen2.5-1.5B-Instruct
 

scripts/eval-mgtv.sh

@@ -1 +1 @@
-eval-mgtv-shots_4bit.sh
+eval-mgtv-qwen2.5.sh