Model	Average	ARC	HellaSwag	MMLU	TruthfulQA
meta-llama/Llama-2-13b-hf	56.9	58.11	80.97	54.34	34.17
meta-llama/Llama-2-13b-chat-hf	59.93	59.04	81.94	54.64	44.12
ehartford/dolphin-llama-13b	59.26	55.55	77.11	52.16	52.23
CHIH-HUNG/llama-2-13b-dolphin_5w	61	60.67	82.69	56.23	44.41
CHIH-HUNG/llama-2-13b-dolphin_20w	60.17	59.56	82.55	55.89	42.67

How to convert dataset to json

在load_dataset中輸入資料集名稱，並且在take中輸入要取前幾筆資料
觀察該資料集的欄位名稱，填入example欄位中(例如instruction、input、output)
最後指定json檔儲存位置 (json_filename)

import json
from datasets import load_dataset

# 讀取數據集，take可以取得該數據集前n筆資料
dataset = load_dataset("ehartford/dolphin", split="train", streaming=True).take(200000)

# 提取所需欄位並建立新的字典列表
extracted_data = []
for example in dataset:
    extracted_example = {
        ### dolphin
        "instruction": example["instruction"],
        "input": example["input"],
        "output": example["output"]
    }
    extracted_data.append(extracted_example)

# 指定 JSON 文件名稱
json_filename = "dolphin.json"

# 寫入 JSON 文件
with open(json_filename, "w") as json_file:
    json.dump(extracted_data, json_file, indent=4)

print(f"數據已提取並保存為 {json_filename}")

CHIH-HUNG/llama-2-13b-dolphin_20w

Model Card for Model ID

Fine-Tuning Information

Fine-Tuning Detail

Evaluation

How to convert dataset to json

CHIH-HUNG/
llama-2-13b-dolphin_20w