Evaluate Codes/2-InternLM2-7b.py

# %%
from transformers import AutoModelForCausalLM, AutoTokenizer
from transformers.generation import GenerationConfig
import torch
import json
import os
from tqdm import tqdm
from functools import partial

tokenizer = AutoTokenizer.from_pretrained("internlm/internlm2-chat-7b", trust_remote_code=True)

model = AutoModelForCausalLM.from_pretrained(
    "internlm/internlm2-chat-7b",
    device_map="cuda:1",
    trust_remote_code=True, torch_dtype=torch.float16
).eval()


def generate_output(file_path, max_new_tokens=128, prefix=""):
    with open(file_path, 'r', encoding='utf-8') as file:
        json_data = json.load(file)
    output = {}
    for index in tqdm(range(len(json_data)), desc=file_path):
        item = json_data[index]
        instruction = item.get("instruction")
        question = item.get("question")
        answer = item.get("answer")

        chat_input = instruction + "\n" + question
        chat_input = chat_input[:2000]
        prediction, history = model.chat(tokenizer, chat_input, history=[], max_new_tokens=64,
                                         do_sample=False, use_cache=True,)
                                         # prefix=prefix)
        output[str(index)] = {
            "origin_prompt": chat_input,
            "prediction": prediction,
            "refr": answer
        }
    return output


folder_path = "../national_test"
output_path = "./InternLM2-7B-Chat"
if not os.path.exists(output_path):
    os.mkdir(output_path)
id_to_task = {
    "4-1": generate_output,
    "4-2": generate_output,
    "4-3": generate_output,
    "4-4": generate_output,
    "4-5": generate_output,
}
for filename in os.listdir(folder_path):
    if filename.endswith('.json'):
        task_name = filename.split(".")[0]
        if task_name not in id_to_task:
            continue
        file_path = os.path.join(folder_path, filename)
        if os.path.exists(os.path.join(output_path, filename)):
            continue
        output = id_to_task[task_name](file_path)
        if not os.path.exists(output_path):
            os.mkdir(output_path)
        with open(os.path.join(output_path, filename), "w", encoding='utf-8') as file:
            json.dump(output, file, ensure_ascii=False)

# %%