Text Generation
Transformers
PyTorch
Safetensors
Vietnamese
English
llama
Eval Results (legacy)
text-generation-inference
Instructions to use capleaf/T-Llama with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use capleaf/T-Llama with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="capleaf/T-Llama")# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("capleaf/T-Llama") model = AutoModelForCausalLM.from_pretrained("capleaf/T-Llama") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use capleaf/T-Llama with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "capleaf/T-Llama" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "capleaf/T-Llama", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/capleaf/T-Llama
- SGLang
How to use capleaf/T-Llama with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "capleaf/T-Llama" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "capleaf/T-Llama", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "capleaf/T-Llama" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "capleaf/T-Llama", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use capleaf/T-Llama with Docker Model Runner:
docker model run hf.co/capleaf/T-Llama
Update README.md
Browse files
README.md
CHANGED
|
@@ -191,6 +191,18 @@ Here's the recommended configuration:
|
|
| 191 |
* penalty_alpha=0.7
|
| 192 |
* repetition_penalty=1.15
|
| 193 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 194 |
Note: 120GB of pre-trained Vietnamese data might not be enough for a general question about Vietnamese events.
|
| 195 |
|
| 196 |
Here is a kaggle script to quickly test the model:
|
|
|
|
| 191 |
* penalty_alpha=0.7
|
| 192 |
* repetition_penalty=1.15
|
| 193 |
|
| 194 |
+
Here is an example using the above generation config:
|
| 195 |
+
```
|
| 196 |
+
> Tôi buồn ngủ quá, cơ mà thứ 7 phải thi rồi, nay là rạng sáng thứ 6. Tôi có nên đi ngủ không trong khi tôi mới xong chưa được 1/3 bài để ôn ?
|
| 197 |
+
|
| 198 |
+
Nếu bạn cảm thấy mệt mỏi và cần nghỉ ngơi thì việc đi ngủ sớm sẽ rất quan trọng đối với sức khỏe của bạn. Tuy nhiên, nếu bạn vẫn muốn hoàn thành một phần bài tập để chuẩn bị cho kỳ thi sắp tới,
|
| 199 |
+
hãy cân nhắc thời gian còn lại của mình và ưu tiên những nhiệm vụ quan trọng nhất trước khi đi ngủ. Nếu bạn cảm thấy rằng mình đã làm đủ, hãy dành chút thời gian thư giãn hoặc thực hiện các
|
| 200 |
+
hoạt động khác để giúp giảm bớt căng thẳng và tăng cường sự tỉnh táo của bạn vào buổi sáng hôm sau. Cuối cùng, quyết định cuối cùng phụ thuộc vào mức độ nghiêm trọng của kỳ thi và khả năng quản
|
| 201 |
+
lý thời gian của bạn.
|
| 202 |
+
|
| 203 |
+
```
|
| 204 |
+
|
| 205 |
+
|
| 206 |
Note: 120GB of pre-trained Vietnamese data might not be enough for a general question about Vietnamese events.
|
| 207 |
|
| 208 |
Here is a kaggle script to quickly test the model:
|