Update README.md

README.md

@@ -39,9 +39,10 @@ from tqdm import tqdm
 import json
 
 # Hugging Faceで取得したTokenをこちらに貼る。
-HF_TOKEN = <YOUR TOKEN>
+from google.colab import userdata
+HF_TOKEN = userdata.get('HF_TOKEN')
 
-# base_model_id = "llm-jp/llm-jp-3-13b"
+model_id = "llm-jp/llm-jp-3-13b"
 adapter_id = "totsukash/llm-jp-3-13b-finetune"
 
 # QLoRA config
@@ -68,7 +69,7 @@ model = PeftModel.from_pretrained(model, adapter_id, token = HF_TOKEN)
 # データセットの読み込み。
 # （評価データセットのjsonlファイルのパスを設定してください）
 datasets = []
-with open("./elyza-tasks-100-TV_0.jsonl", "r") as f:
+with open("/content/elyza-tasks-100-TV_0.jsonl", "r") as f:
     item = ""
     for line in f:
       line = line.strip()
@@ -80,18 +81,18 @@ with open("./elyza-tasks-100-TV_0.jsonl", "r") as f:
 # gemma
 results = []
 for data in tqdm(datasets):
-
-  input = data["input"]
-  prompt = f"""### 指示
-  {input}
-  ### 回答
-  """
-
-  input_ids = tokenizer(prompt, return_tensors="pt").to(model.device)
-  outputs = model.generate(**input_ids, max_new_tokens=512, do_sample=False, repetition_penalty=1.2,)
-  output = tokenizer.decode(outputs[0][input_ids.input_ids.size(1):], skip_special_tokens=True)
-
-  results.append({"task_id": data["task_id"], "input": input, "output": output})
+    input = data["input"]
+    prompt = f"""### 指示
+    {input}
+    ### 回答
+    """
+    
+    # input_ids だけを取り出して使用
+    input_ids = tokenizer(prompt, return_tensors="pt").input_ids.to(model.device)
+    outputs = model.generate(input_ids, max_new_tokens=512, do_sample=False, repetition_penalty=1.2)
+    output = tokenizer.decode(outputs[0][input_ids.size(1):], skip_special_tokens=True)
+    
+    results.append({"task_id": data["task_id"], "input": input, "output": output})
 
 # llmjp
 results = []