Update README.md

README.md

@@ -8,19 +8,18 @@ tags:
 
 # MedSwallow-70B🏥
 
-[東工大Swallow](tokyotech-llm/Swallow-70b-instruct-hf)をベースモデルとし, 医療Q&AデータセットでInstruction Tuningを施した医療ドメインの日本語LLMです.
+[東工大Swallow](https://huggingface.co/tokyotech-llm/Swallow-70b-instruct-hf)をベースモデルとし, 医療Q&AデータセットでInstruction Tuningを施した医療ドメインの日本語LLMです.
 
 チューニングには独自で用意した米国医師国家試験(USMLE)を和訳したQ&Aデータセットを用いました.
 
 
 MedSwallow is a Japanese medical LLM for medical question-answering.
 
-MedSwallow is based on [Swallow-70B]((tokyotech-llm/Swallow-70b-instruct-hf)) and has passed instruction tuning with USMLE dataset translated in Japanese by our own.
+MedSwallow is based on [Swallow-70B](https://huggingface.co/tokyotech-llm/Swallow-70b-instruct-hf) and has passed instruction tuning with USMLE dataset translated in Japanese by our own.
 
 
 ## Training procedure
 
-
 The following `bitsandbytes` quantization config was used during training:
 - quant_method: bitsandbytes
 - load_in_8bit: False
@@ -44,6 +43,46 @@ The following `bitsandbytes` quantization config was used during training:
 Non-commercial.
 
 
+## Usage
+
+```
+model_name = "tokyotech-llm/Swallow-70b-instruct-hf"
+peft_model= "AIgroup-CVM-utokyohospital/MedSwallow-70b"
+
+tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
+
+bnb_config = BitsAndBytesConfig(
+            load_in_4bit=True,
+            bnb_4bit_quant_type="nf4",
+            bnb_4bit_compute_dtype=torch.float16,
+        )
+
+model = AutoModelForCausalLM.from_pretrained(
+    model_name,
+    load_in_8bit=False,
+    torch_dtype=torch.float16,
+    device_map=device,
+        
+model = PeftModel.from_pretrained(
+    model, 
+    peft_model, 
+    torch_dtype=torch.float16,
+    device_map=device, 
+)
+
+```
+
+
+## Benchmark
+
+See also [Japanese Medical Language Model Evaluation Harness](https://github.com/stardust-coder/japanese-lm-med-harness).
+
+- IgakuQA (in English): 
+- IgakuQA (in Japanese): 
+- MedQA (in English) :
+- MedQA (in Japanese) :
+
+
 ## How to cite
 ```
 coming soon...