Update README.md

README.md

@@ -14,6 +14,27 @@ pipeline_tag: text-generation
 **Adapter Training Corpus**
 - [maywell/ko_Ultrafeedback_binarized](https://huggingface.co/datasets/maywell/ko_Ultrafeedback_binarized)
 
+**Explanation**
+- With base model, merge the saved DPO applied adapter
+
+**Score**
+|Average|ARC|HellaSwag|MMLU|TruthfulQA|Winogrande|GSM8K|
+|:---:|:---:|:---:|:---:|:---:|:---:|:---:|
+|74.11|71.08|88.2|66.09|71.51|83.5|64.29|
+
+**Usage**
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
+model = AutoModelForCausalLM.from_pretrained(
+        "dddsaty/Open_Ko_SOLAR_DPO_Merge_v0.1",
+        low_cpu_mem_usage = True,
+        torch_dtype = torch.float16,
+        device_map = device_map,
+    )
+
+tokenizer = AutoTokenizer.from_pretrained("dddsaty/Open_Ko_SOLAR_DPO_Merge_v0.1")
+```
+
 **Log**
 - 2024.01.29: Initial version Upload
 
@@ -21,11 +42,6 @@ pipeline_tag: text-generation
 Following the upstage/SOLAR-10.7B-Instruct-v1.0 License
 - cc-by-nc-4.0
 
-**Score**
-|Average|ARC|HellaSwag|MMLU|TruthfulQA|Winogrande|GSM8K|
-|:---:|:---:|:---:|:---:|:---:|:---:|:---:|
-|74.11|71.08|88.2|66.09|71.51|83.5|64.29|
-
 **Citation**
 - beomi/OPEN-SOLAR-KO-10.7B  
 ```