Downloads · 30 days
0
Hikaru-Yamashita0804/llm-jp-3-13b-finetune-2
llm-jp-3-13b-finetune-2 is a machine learning model from Hikaru-Yamashita0804. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
Downloads · 30 days
0
Access
Public
Updated Nov 25, 2024
Repo size
501 MB
Likes
0
Public
Click a slice to open those files.
.safetensors501 MB · 99%
From the Hugging Face model README
・Developed by: YHikaru413879
・Finetuned from model : llm-jp/llm-jp-3-13b
・License: apache-2.0
・学習データセット Ichikara Instruction_Distribution20241221_all のほぼ全てのデータセット
# llm-jp/llm-jp-3-13bを4bit量子化のqLoRA設定でロード。
from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
from unsloth import FastLanguageModel
import torch
max_seq_length = 512 # unslothではRoPEをサポートしているのでコンテキスト長は自由に設定可能
dtype = None # Noneにしておけば自動で設定
load_in_4bit = True # 今回は8Bクラスのモデルを扱うためTrue
model_id = "llm-jp/llm-jp-3-13b"
new_model_id = "llm-jp-3-13b-finetune-2" #Fine-Tuningしたモデルにつけたい名前
# FastLanguageModel インスタンスを作成
model, tokenizer = FastLanguageModel.from_pretrained(
model_name=model_id,
dtype=dtype,
load_in_4bit=load_in_4bit,
trust_remote_code=True,
)
# SFT用のモデルを用意
model = FastLanguageModel.get_peft_model(
model,
r = 32,
target_modules = ["q_proj", "k_proj", "v_proj", "o_proj",
"gate_proj", "up_proj", "down_proj",],
lora_alpha = 32,
lora_dropout = 0.05,
bias = "none",
use_gradient_checkpointing = "unsloth",
random_state = 3407,
use_rslora = False,
loftq_config = None,
max_seq_length = max_seq_length,
)
# 学習時のプロンプトフォーマットの定義
prompt = """### 指示
{}
### 回答
{}"""
"""
formatting_prompts_func: 各データをプロンプトに合わせた形式に合わせる
"""
EOS_TOKEN = tokenizer.eos_token # トークナイザーのEOSトークン(文末トークン)
def formatting_prompts_func(examples):
input = examples["text"] # 入力データ
output = examples["output"] # 出力データ
text = prompt.format(input, output) + EOS_TOKEN # プロンプトの作成
return { "formatted_text" : text, } # 新しいフィールド "formatted_text" を返す
pass
# # 各データにフォーマットを適用
dataset = dataset.map(
formatting_prompts_func,
num_proc= 4, # 並列処理数を指定
)
dataset
from trl import SFTTrainer
from transformers import TrainingArguments
from unsloth import is_bfloat16_supported
trainer = SFTTrainer(
model = model,
tokenizer = tokenizer,
train_dataset=dataset["train"],
max_seq_length = max_seq_length,
dataset_text_field="formatted_text",
packing = False,
args = TrainingArguments(
per_device_train_batch_size = 2,
gradient_accumulation_steps = 4,
num_train_epochs = 1,
logging_steps = 10,
warmup_steps = 10,
save_steps=100,
save_total_limit=2,
max_steps=-1,
learning_rate = 2e-4,
fp16 = not is_bfloat16_supported(),
bf16 = is_bfloat16_supported(),
group_by_length=True,
seed = 3407,
output_dir = "outputs",
report_to = "none",
),
)
#@title 学習実行
trainer_stats = trainer.train()
# 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})
base_model: llm-jp/llm-jp-3-13b tags:
This llama model was trained 2x faster with Unsloth and Huggingface's TRL library.