metadata
library_name: transformers
datasets:
- elyza/ELYZA-tasks-100
license: apache-2.0
language:
- ja
base_model:
- llm-jp/llm-jp-3-13b-instruct
Model Card for Model ID
Required Libraries and Their Versions
- trl==0.12.2
- transformers<4.47.0
- tokenizers==0.21.0
Usage
results = []
system_text = "以下は、タスクを説明する指示です。要求を適切に満たす回答を**簡潔に**書いてください。回答の後ろに、回答の理由を**1文で**書いてください。"
for data in tqdm(datasets):
input_text = data["input"]
prompt = f"""
{system_text}
### 指示
{input_text}
### 応答
"""
tokenized_input = tokenizer.encode(prompt, add_special_tokens=False, return_tensors="pt").to(model.device)
attention_mask = torch.ones_like(tokenized_input)
with torch.no_grad():
outputs = model.generate(
tokenized_input,
attention_mask=attention_mask,
max_new_tokens=100,
do_sample=False,
repetition_penalty=1.2,
pad_token_id=tokenizer.eos_token_id
)[0]
output = tokenizer.decode(outputs[tokenized_input.size(1):], skip_special_tokens=True)
results.append({"task_id": data["task_id"], "input": input_text, "output": output})
Model Details
- Model type: Transformer-based Language Model
Datasets
Instruction tuning
| Language | Dataset | description |
|---|---|---|
| Japanese | elyza/ELYZA-tasks-100 | A manually constructed instruction dataset |