Instructions to use hibikaze/tiny_mixtral_ja_with_tokenizer with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use hibikaze/tiny_mixtral_ja_with_tokenizer with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="hibikaze/tiny_mixtral_ja_with_tokenizer")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("hibikaze/tiny_mixtral_ja_with_tokenizer")
model = AutoModelForCausalLM.from_pretrained("hibikaze/tiny_mixtral_ja_with_tokenizer")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use hibikaze/tiny_mixtral_ja_with_tokenizer with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "hibikaze/tiny_mixtral_ja_with_tokenizer"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "hibikaze/tiny_mixtral_ja_with_tokenizer",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/hibikaze/tiny_mixtral_ja_with_tokenizer

SGLang

How to use hibikaze/tiny_mixtral_ja_with_tokenizer 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 "hibikaze/tiny_mixtral_ja_with_tokenizer" \
    --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": "hibikaze/tiny_mixtral_ja_with_tokenizer",
		"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 "hibikaze/tiny_mixtral_ja_with_tokenizer" \
        --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": "hibikaze/tiny_mixtral_ja_with_tokenizer",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use hibikaze/tiny_mixtral_ja_with_tokenizer with Docker Model Runner:
```
docker model run hf.co/hibikaze/tiny_mixtral_ja_with_tokenizer
```

275.86Mのmixtralを日本語データセットでpretrainingしたものです

sample

from transformers import AutoTokenizer, AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("if001/tiny_mixtral_ja")
tokenizer = AutoTokenizer.from_pretrained("if001/sentencepiece_ja", trust_remote_code=True)

prompt = "それは九月初旬のある蒸し暑い晩のことであった。私は、Ｄ坂の"
inputs = tokenizer(prompt, return_tensors="pt")
 
generate_ids = model.generate(
    inputs.input_ids, 
    max_length=30,
    top_k=30,
    top_p=0.95,
    temperature=0.6,
    repetition_penalty=1.2,
    do_sample=True,
)
tokenizer.decode(generate_ids[0], skip_special_tokens=True, clean_up_tokenization_spaces=False)

>> それは九月初旬のある蒸し暑い晩のことであった。私は、D坂の茶舗を後にして、その路地の角に横丁をあるいて居る、と云うと、丁度其処から、

dataset

英語と日本語のデータセットを使用

total tokens: 8.64B

wikipedia_ja:    844.65M  
wikipedia_en:    3.80B  
open-text-books: 60.17M  
oscar:           3.85B  
aozorabunko:     92.97M

tokenizer

all_special_ids:  [1, 2, 3, 0, 4]
all_special_tokens:  ['<BOS>', '<EOS>', '<UNK>', '<PAD>', '<MASK>']

Downloads last month: 9

Safetensors

Model size

0.3B params

Tensor type

F32

hibikaze
/

tiny_mixtral_ja_with_tokenizer

sample

dataset

tokenizer

Datasets used to train hibikaze/tiny_mixtral_ja_with_tokenizer