pip install unsloth
pip uninstall unsloth -y && pip install --upgrade --no-cache-dir "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
pip install -U torch
pip install -U peft

code

from unsloth import FastLanguageModel
from peft import PeftModel
import torch
import json
from tqdm import tqdm
import re
model_id = "llm-jp/llm-jp-3-13b"
adapter_id = "baxin/llm-jp-3-13b-it-seq-r-la_lora_test"

HF_TOKEN=""

dtype = None 
load_in_4bit = True

model, tokenizer = FastLanguageModel.from_pretrained(
    model_name=model_id,
    dtype=dtype,
    load_in_4bit=load_in_4bit,
    trust_remote_code=True,
)

model = PeftModel.from_pretrained(model, adapter_id, token = HF_TOKEN)

# load elyza-tasks-100-TV jsonl
datasets = []
with open("./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 = ""

FastLanguageModel.for_inference(model)

system_prompt = "あなたは指示に注意深く従う親切なアシスタントです。指示をステップバイステップで理解し、回答してください。"

results = []
for dt in tqdm(datasets):
  input = dt["input"]

  prompt = f"""{system_prompt}\n\n### 指示\n{input}\n### 回答\n"""

  inputs = tokenizer([prompt], return_tensors = "pt").to(model.device)

  outputs = model.generate(**inputs, max_new_tokens = 3072, 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})

Uploaded model

Developed by: baxin
License: apache-2.0
Finetuned from model : llm-jp/llm-jp-3-13b

This llama model was trained 2x faster with Unsloth and Huggingface's TRL library.