From f15e37dfad640378836930d640241a630182b82e Mon Sep 17 00:00:00 2001
From: wql <wengqinlan18@163.com>
Date: Thu, 5 Sep 2024 15:49:32 +0800
Subject: [PATCH] fix: fix bf16

---
 prepare_yaml_file.py           | 2 +-
 results/lora_sft_template.yaml | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/prepare_yaml_file.py b/prepare_yaml_file.py
index 7428f229..36357f6b 100644
--- a/prepare_yaml_file.py
+++ b/prepare_yaml_file.py
@@ -28,7 +28,7 @@ def main():
         model_name_or_path = "../../models/llama-2-7b-ms"
         template = "llama2"
     elif model == "Qwen-7B":
-        model_name_or_path = "/root/models/Qwen-7B"
+        model_name_or_path = "../../models/Qwen-7B"
         template = "qwen"
     else:
         print("ERROR: model not supported.")
diff --git a/results/lora_sft_template.yaml b/results/lora_sft_template.yaml
index a3b42642..9a4411e4 100644
--- a/results/lora_sft_template.yaml
+++ b/results/lora_sft_template.yaml
@@ -29,7 +29,7 @@ learning_rate: 1.0e-4
 num_train_epochs: 10.0
 lr_scheduler_type: cosine
 warmup_ratio: 0.1
-bf16: true
+fp16: true
 ddp_timeout: 180000000
 max_steps: 500
 include_num_input_tokens_seen: true