Update README.md
Browse files
README.md
CHANGED
|
@@ -16,8 +16,10 @@ license: apache-2.0
|
|
| 16 |
This model learned the proceedings of the Japanese parliament in 2022.
|
| 17 |
The [dataset](https://huggingface.co/datasets/ohtaman/kokkai2022) is collected using
|
| 18 |
[National Diet Library's Search API](https://kokkai.ndl.go.jp/api.html).
|
|
|
|
|
|
|
| 19 |
|
| 20 |
-
example input:
|
| 21 |
|
| 22 |
```
|
| 23 |
# question
|
|
@@ -28,7 +30,7 @@ example input:
|
|
| 28 |
鈴木 俊一
|
| 29 |
```
|
| 30 |
|
| 31 |
-
|
| 32 |
|
| 33 |
```
|
| 34 |
「財政民主主義」のためには、国庫負担を引き下げるならば、企業の賃上げを実現するためにも、消費者物価の高騰対策等を含めて、経済対策を行い、成長と分配の好循環を実珉化することが重要でございます。
|
|
@@ -82,7 +84,7 @@ base_model = transformers.AutoModelForCausalLM.from_pretrained(base_model_name,
|
|
| 82 |
peft_model = peft.PeftModelForCausalLM.from_pretrained(base_model, peft_model_name, torch_dtype=torch.bfloat16)
|
| 83 |
|
| 84 |
|
| 85 |
-
prompt = "# question\n麻生太郎\n\n
|
| 86 |
input_tokens = tokenizer(prompt, return_tensors="pt").to(peft_model.device)
|
| 87 |
input_length = input_tokens.input_ids.shape[1]
|
| 88 |
|
|
|
|
| 16 |
This model learned the proceedings of the Japanese parliament in 2022.
|
| 17 |
The [dataset](https://huggingface.co/datasets/ohtaman/kokkai2022) is collected using
|
| 18 |
[National Diet Library's Search API](https://kokkai.ndl.go.jp/api.html).
|
| 19 |
+
This model was build for a hackerthon event, [
|
| 20 |
+
第1回大規模言語モデル分散学習ハッカソン](https://abci.ai/event/2023/06/13/ja_event.html) ([#ABCILLM](https://twitter.com/hashtag/ABCILLM)), as an example of training which used multiple GPUs or multiple nodes.
|
| 21 |
|
| 22 |
+
An example input is as follows:
|
| 23 |
|
| 24 |
```
|
| 25 |
# question
|
|
|
|
| 30 |
鈴木 俊一
|
| 31 |
```
|
| 32 |
|
| 33 |
+
and the respons is:
|
| 34 |
|
| 35 |
```
|
| 36 |
「財政民主主義」のためには、国庫負担を引き下げるならば、企業の賃上げを実現するためにも、消費者物価の高騰対策等を含めて、経済対策を行い、成長と分配の好循環を実珉化することが重要でございます。
|
|
|
|
| 84 |
peft_model = peft.PeftModelForCausalLM.from_pretrained(base_model, peft_model_name, torch_dtype=torch.bfloat16)
|
| 85 |
|
| 86 |
|
| 87 |
+
prompt = "# question\n麻生太郎\n\n増税が必要とお考えでしょうか？\n# answer\n鈴木 俊一\n\n"
|
| 88 |
input_tokens = tokenizer(prompt, return_tensors="pt").to(peft_model.device)
|
| 89 |
input_length = input_tokens.input_ids.shape[1]
|
| 90 |
|