Instructions to use hillfieldhonest/aozoranemo292m with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use hillfieldhonest/aozoranemo292m with Transformers:

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

pipe = pipeline("text-generation", model="hillfieldhonest/aozoranemo292m")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("hillfieldhonest/aozoranemo292m")
model = AutoModelForCausalLM.from_pretrained("hillfieldhonest/aozoranemo292m", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use hillfieldhonest/aozoranemo292m with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "hillfieldhonest/aozoranemo292m"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "hillfieldhonest/aozoranemo292m",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/hillfieldhonest/aozoranemo292m

SGLang

How to use hillfieldhonest/aozoranemo292m 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 "hillfieldhonest/aozoranemo292m" \
    --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": "hillfieldhonest/aozoranemo292m",
		"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 "hillfieldhonest/aozoranemo292m" \
        --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": "hillfieldhonest/aozoranemo292m",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use hillfieldhonest/aozoranemo292m with Docker Model Runner:
```
docker model run hf.co/hillfieldhonest/aozoranemo292m
```

aozoranemo292m

Japanese mini LLM (~292M params). SentencePiece 16k.

Parameters: 292,478,784
Tokenizer: SentencePiece 16k
Best checkpoint (logs): checkpoint-100（継続学習は step=600 まで）

Training data (references)

Aozora Bunko (clean): https://huggingface.co/datasets/globis-university/aozorabunko-clean
Nemotron Personas (Japan): https://huggingface.co/blog/nvidia/nemotron-personas-japan?linkId=100000383918769 ※ 各データのライセンスに従ってご利用ください。

Quick start

from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
repo = "hillfieldhonest/aozoranemo292m"
tok = AutoTokenizer.from_pretrained(repo, use_fast=False)
if tok.pad_token is None: tok.pad_token = tok.eos_token
dtype = (torch.bfloat16 if torch.cuda.is_available() and torch.cuda.is_bf16_supported()
         else torch.float16 if torch.cuda.is_available() else torch.float32)
model = AutoModelForCausalLM.from_pretrained(repo, dtype=dtype, device_map="auto").eval()
out = model.generate(**tok("2文以内で自己紹介して。", return_tensors="pt").to(model.device),
                     max_new_tokens=96, do_sample=True, temperature=0.7, top_p=0.9)
print(tok.decode(out[0], skip_special_tokens=True))

License

Weights: Apache-2.0（暫定。必要に応じて調整）
Data: 上記参照先の規定に準拠

Downloads last month: 3

Safetensors

Model size

0.3B params

Tensor type

F32