README.md

---
base_model: google/gemma-2-9b
tags:
- text-generation-inference
- transformers
- unsloth
- gemma2
- trl
license: cc-by-sa-4.0
language:
- en
- ja
---

# Uploaded  model

- **Developed by:** Kohsaku
- **License:** CC BY-NC-ND 4.0
- **Finetuned from model :** google/gemma-2-9b

This gemma2 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)

推論コード

なお、環境変数 HF_TOKENは別途設定されているものとします。

``` pip install
# Colabratory例
!pip uninstall unsloth -y
!pip install --upgrade --no-cache-dir "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
!pip install --upgrade torch
!pip install --upgrade xformers
!pip install ipywidgets --upgrade

import torch
if torch.cuda.get_device_capability()[0] >= 8:
    !pip install --no-deps packaging ninja einops "flash-attn>=2.6.3"
```

``` sample code
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
from unsloth import FastLanguageModel
import torch
import json

model_name = "Kohsaku/gemma-2-9b-finetune-4"

max_seq_length = 1024

dtype = None
load_in_4bit = True

model, tokenizer = FastLanguageModel.from_pretrained(
    model_name = model_name,
    max_seq_length = max_seq_length,
    dtype = dtype,
    load_in_4bit = load_in_4bit,
    token = HF_TOKEN,
)
FastLanguageModel.for_inference(model)

text = "自然言語処理とは何か"
tokenized_input = tokenizer.encode(text, add_special_tokens=True , return_tensors="pt").to(model.device)

with torch.no_grad():
    output = model.generate(
        tokenized_input,
        max_new_tokens = 1024,
        use_cache = True,
        do_sample=False,
        repetition_penalty=1.2
    )[0]
print(tokenizer.decode(output))

# ELYZA-tasks-100-TVによる評価
# ELYZA-tasks-100-TVの読み込み。事前にファイルをアップロードしてください
# データセットの読み込み。
# omnicampusの開発環境では、左にタスクのjsonlをドラッグアンドドロップしてから実行。
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 = ""

# 学習したモデルを用いてタスクを実行
from tqdm import tqdm

# 推論するためにモデルのモードを変更
FastLanguageModel.for_inference(model)

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

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

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

  outputs = model.generate(**inputs, max_new_tokens = 512, 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})

# jsonlで保存
with open(f"{model_name.split('/')[-1]}_outputs.jsonl", 'w', encoding='utf-8') as f:
    for result in results:
        json.dump(result, f, ensure_ascii=False)
        f.write('\n')
```