Update README.md
Browse files
README.md
CHANGED
|
@@ -14,7 +14,7 @@ language:
|
|
| 14 |
# Uploaded model
|
| 15 |
|
| 16 |
- **Developed by:** umizkimt
|
| 17 |
-
- **License:**
|
| 18 |
- **Finetuned from model :** llm-jp/llm-jp-3-13b
|
| 19 |
|
| 20 |
This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
|
|
@@ -44,7 +44,9 @@ This llama model was trained 2x faster with [Unsloth](https://github.com/unsloth
|
|
| 44 |
## Performance
|
| 45 |
- Omnicampus score: 3.02 (2024-11-29 19:20:27 JST)
|
| 46 |
|
| 47 |
-
##
|
|
|
|
|
|
|
| 48 |
```python
|
| 49 |
# 必要なライブラリをインストール
|
| 50 |
%%capture
|
|
@@ -52,10 +54,12 @@ This llama model was trained 2x faster with [Unsloth](https://github.com/unsloth
|
|
| 52 |
!pip uninstall unsloth -y && pip install --upgrade --no-cache-dir "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
|
| 53 |
!pip install -U torch
|
| 54 |
!pip install -U peft
|
| 55 |
-
|
|
|
|
| 56 |
# unsloth: 2024.11.10
|
| 57 |
# peft: 0.13.2
|
| 58 |
# torch: 2.5.1+cu121
|
|
|
|
| 59 |
|
| 60 |
# 必要なライブラリを読み込み
|
| 61 |
from unsloth import FastLanguageModel
|
|
@@ -70,7 +74,7 @@ model_id = "llm-jp/llm-jp-3-13b"
|
|
| 70 |
adapter_id = "umizkimt/llm-jp-3-13b-it_lora"
|
| 71 |
|
| 72 |
# Hugging Face Token を指定。
|
| 73 |
-
HF_TOKEN = "<your-
|
| 74 |
|
| 75 |
# unslothのFastLanguageModelで元のモデルをロード。
|
| 76 |
dtype = None # Noneにしておけば自動で設定
|
|
@@ -126,12 +130,15 @@ with open(f"/content/{json_file_id}_output.jsonl", 'w', encoding='utf-8') as f:
|
|
| 126 |
f.write('\n')
|
| 127 |
```
|
| 128 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 129 |
## Limitaitons
|
| 130 |
This model is in early development stages. Outputs may not consistently align with human intent and require careful validation. Potential for generating inappropriate or incorrect responses exists. Recommended for experimental use with human oversight.
|
| 131 |
|
| 132 |
## License
|
| 133 |
-
TBD
|
| 134 |
-
|
| 135 |
This fine-tuned model is released under the CC BY-NC-SA 4.0 license, as it was trained on a dataset covered by the same license. The pre-trained model used as a starting point for fine-tuning is distributed under the Apache License 2.0.
|
| 136 |
|
| 137 |
## Model Card Authors
|
|
|
|
| 14 |
# Uploaded model
|
| 15 |
|
| 16 |
- **Developed by:** umizkimt
|
| 17 |
+
- **License:** CC BY-NC-SA 4.0
|
| 18 |
- **Finetuned from model :** llm-jp/llm-jp-3-13b
|
| 19 |
|
| 20 |
This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
|
|
|
|
| 44 |
## Performance
|
| 45 |
- Omnicampus score: 3.02 (2024-11-29 19:20:27 JST)
|
| 46 |
|
| 47 |
+
## .jsonl File Output Usage
|
| 48 |
+
To generate the output file in Google Colaboratory, use the following script:
|
| 49 |
+
|
| 50 |
```python
|
| 51 |
# 必要なライブラリをインストール
|
| 52 |
%%capture
|
|
|
|
| 54 |
!pip uninstall unsloth -y && pip install --upgrade --no-cache-dir "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
|
| 55 |
!pip install -U torch
|
| 56 |
!pip install -U peft
|
| 57 |
+
|
| 58 |
+
# Omnicampusに提出した.jsonlファイル出力時のライブラリバージョン
|
| 59 |
# unsloth: 2024.11.10
|
| 60 |
# peft: 0.13.2
|
| 61 |
# torch: 2.5.1+cu121
|
| 62 |
+
# tqdm: 4.66.6
|
| 63 |
|
| 64 |
# 必要なライブラリを読み込み
|
| 65 |
from unsloth import FastLanguageModel
|
|
|
|
| 74 |
adapter_id = "umizkimt/llm-jp-3-13b-it_lora"
|
| 75 |
|
| 76 |
# Hugging Face Token を指定。
|
| 77 |
+
HF_TOKEN = "<your-token>"
|
| 78 |
|
| 79 |
# unslothのFastLanguageModelで元のモデルをロード。
|
| 80 |
dtype = None # Noneにしておけば自動で設定
|
|
|
|
| 130 |
f.write('\n')
|
| 131 |
```
|
| 132 |
|
| 133 |
+
### Execution Details
|
| 134 |
+
- Platform: Google Colaboratory
|
| 135 |
+
- Hardware: T4 GPU
|
| 136 |
+
- Approximate Inference Time: 9 minutes
|
| 137 |
+
|
| 138 |
## Limitaitons
|
| 139 |
This model is in early development stages. Outputs may not consistently align with human intent and require careful validation. Potential for generating inappropriate or incorrect responses exists. Recommended for experimental use with human oversight.
|
| 140 |
|
| 141 |
## License
|
|
|
|
|
|
|
| 142 |
This fine-tuned model is released under the CC BY-NC-SA 4.0 license, as it was trained on a dataset covered by the same license. The pre-trained model used as a starting point for fine-tuning is distributed under the Apache License 2.0.
|
| 143 |
|
| 144 |
## Model Card Authors
|