From fad73a77f1b11e32f0d14c8ccf9ce23883670af4 Mon Sep 17 00:00:00 2001
From: lwaekfjlk <1125027232@qq.com>
Date: Wed, 11 Oct 2023 02:31:27 +0000
Subject: [PATCH] add qlora

---
 llm_ft/data/dummyfile |  0
 llm_ft/dummy_qlora.sh | 25 +++++++++++++++++++++++++
 2 files changed, 25 insertions(+)
 create mode 100644 llm_ft/data/dummyfile
 create mode 100644 llm_ft/dummy_qlora.sh

diff --git a/llm_ft/data/dummyfile b/llm_ft/data/dummyfile
new file mode 100644
index 00000000..e69de29b
diff --git a/llm_ft/dummy_qlora.sh b/llm_ft/dummy_qlora.sh
new file mode 100644
index 00000000..ddd9a752
--- /dev/null
+++ b/llm_ft/dummy_qlora.sh
@@ -0,0 +1,25 @@
+deepspeed fastchat/train/train_lora.py \
+    --model_name_or_path ./vicuna-7b-1.5 \
+    --lora_r 8 \
+    --lora_alpha 16 \
+    --lora_dropout 0.05 \
+    --data_path ./data/dummy_conversation.json \
+    --bf16 True \
+    --output_dir ./checkpoints \
+    --num_train_epochs 3 \
+    --per_device_train_batch_size 1 \
+    --per_device_eval_batch_size 1 \
+    --gradient_accumulation_steps 1 \
+    --evaluation_strategy "no" \
+    --save_strategy "steps" \
+    --save_steps 1200 \
+    --save_total_limit 100 \
+    --learning_rate 2e-5 \
+    --weight_decay 0. \
+    --warmup_ratio 0.03 \
+    --lr_scheduler_type "cosine" \
+    --logging_steps 1 \
+    --tf32 True \
+    --model_max_length 2048 \
+    --q_lora True \
+    --deepspeed playground/deepspeed_config_s2.json
\ No newline at end of file