File size: 2,977 Bytes
a108004 4e5f73a a108004 4e5f73a a108004 4e5f73a a108004 635bd2a 325b999 635bd2a 4e5f73a 325b999 4e5f73a 0451c76 4e5f73a 0451c76 4e5f73a e781167 4e5f73a a108004 4e5f73a 325b999 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 |
---
base_model: llm-jp/llm-jp-3-13b
license: apache-2.0
language:
- ja
datasets:
- kajuma/dpo_1
---
# Model Card for JunichiroMorita/llm-jp-3-13b-it_lora_20241216
## Model Details
- **Developed by:** JunichiroMorita
- **Language(s) (NLP):** Japanese
- **License:** Apache license 2.0
- **Finetuned from model :** llm-jp/llm-jp-3-13b
## Description
This model was developed for use in a competition, specifically for [松尾研大規模言語モデル講座2024](https://weblab.t.u-tokyo.ac.jp/lecture/course-list/large-language-model/).
## Uses
```python
!pip install unsloth
!pip install --upgrade --no-cache-dir "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
!pip install -U torch
!pip install -U peft
```
```python
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 = f"JunichiroMorita/llm-jp-3-13b-it_lora_20241216"
HF_TOKEN = 'your_hugging_face_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)
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)
results = []
for dt in tqdm(datasets):
input = dt["input"]
prompt = f"""### 指示\n{input}\n\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\n')[-1]
results.append({"task_id": dt["task_id"], "input": input, "output": prediction})
with open(f'./llm-jp-3-13b-it_lora_20241216_output.jsonl', 'w', encoding='utf-8') as f:
for result in results:
json.dump(result, f, ensure_ascii=False)
f.write('\n')
```
## Training Details
### Training Data
- [kajuma/dpo_1](https://huggingface.co/datasets/kajuma/dpo_1)
### Training Procedure
This model was fine-tuned using LoRA (Low-Rank Adaptation) to optimize training efficiency and minimize computational overhead while maintaining performance. The fine-tuning process utilized Japanese instruction data specifically designed for LLMs to enhance its capabilities in understanding and generating Japanese-language instructions.
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)
|