Update README.md

README.md

@@ -23,8 +23,9 @@ This llama model was trained 2x faster with [Unsloth](https://github.com/unsloth
 
 ---
 
-#Google Colabでの動作を想定
+# Google Colabでの動作を想定
 
+```python
 # 必要なライブラリをインストール
 %%capture
 !pip install unsloth
@@ -59,7 +60,7 @@ model, tokenizer = FastLanguageModel.from_pretrained(
 )
 
 # 元のモデルにLoRAのアダプタを統合
-model = PeftModel.from_pretrained(model, adapter_id, token = HF_TOKEN)
+model = PeftModel.from_pretrained(model, adapter_id, token=HF_TOKEN)
 
 # google drive mount(事前にデータをアップロード)
 from google.colab import drive
@@ -70,26 +71,27 @@ datasets = []
 with open("/content/drive/MyDrive/2024松尾研LLM/elyza-tasks-100-TV_0.jsonl", "r") as f:
     item = ""
     for line in f:
-      line = line.strip()
-      item += line
-      if item.endswith("}"):
-        datasets.append(json.loads(item))
-        item = ""
+        line = line.strip()
+        item += line
+        if item.endswith("}"):
+            datasets.append(json.loads(item))
+            item = ""
+
 
 # モデルを用いてタスクの推論
 FastLanguageModel.for_inference(model)
 
 results = []
 for dt in tqdm(datasets):
-  input = dt["input"]
+    input = dt["input"]
+
+    prompt = f"""### 指示\n{input}\n### 回答\n"""
 
-  prompt = f"""### 指示\n{input}\n### 回答\n"""
+    inputs = tokenizer([prompt], return_tensors="pt").to(model.device)
 
-  inputs = tokenizer([prompt], return_tensors = "pt").to(model.device)
+    outputs = model.generate(**inputs, max_new_tokens=512, use_cache=True, do_sample=False, repetition_penalty=1.2)
+    prediction = tokenizer.decode(outputs[0], skip_special_tokens=True).split('\n### 回答')[-1]
 
-  outputs = model.generate(**inputs, max_new_tokens = 512, use_cache = True, do_sample=False, repetition_penalty=1.2)
-  prediction = tokenizer.decode(outputs[0], skip_special_tokens=True).split('\n### 回答')[-1]
+    results.append({"task_id": dt["task_id"], "input": input, "output": prediction})
 
-  results.append({"task_id": dt["task_id"], "input": input, "output": prediction})
-  
-  ---
+---