---
base_model: llm-jp/llm-jp-3-13b
tags:
- text-generation-inference
- transformers
- unsloth
- llama
- trl
license: apache-2.0
language:
- ja
datasets:
- msfm/ichikara-instruction-all
- elyza/ELYZA-tasks-100
---

# Uploaded  model

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

This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.

[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)


## 出力方法
```python
from unsloth import FastLanguageModel
from peft import PeftModel
from tqdm import tqdm
import torch
import json
import re


#モデルのロード
max_seq_length = 1024
dtype = None
load_in_4bit = True

model_id = "llm-jp/llm-jp-3-13b"

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


#RoLAアダプタのロード
adapter_id = "satami/llm-jp-3-13b_LoRA_with_policy_v2"
HF_TOKEN = "your_token"

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


#データセットのロード
datasets = []
with open("../../dataset/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 = ""


#タスクの実行
FastLanguageModel.for_inference(model)
results = []
for dt in tqdm(datasets):
  input = dt["input"]

  prompt = f"""以下は、タスクを説明する指示です。自分が何を出力するかを簡潔に答えた後、要求を適切に満たす応答を書きなさい。

### 指示:{input}

### タスク:"""

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

  outputs = model.generate(**inputs, max_new_tokens = 1024, use_cache = True, do_sample=False, repetition_penalty=1.2)
  prediction = tokenizer.decode(outputs[0], skip_special_tokens=True).split('\n### タスク:')[-1]

  results.append({"task_id": dt["task_id"], "input": input, "output": prediction})


#本モデルはタスクの要約と応答を出力するので、タスクの要約を出力から削除します。
for i in range(len(results)):
    results[i]["output"] = results[i]["output"].split("### 応答:")[-1]


# jsonlで保存
new_model_id = "ichikara-elyza_r=16_with-policy"
with open(f"{new_model_id}_output.jsonl", 'w', encoding='utf-8') as f:
    for result in results:
        json.dump(result, f, ensure_ascii=False)
        f.write('\n')
```