Instructions to use Ismantic/Qwen3-1.7B-Base-ReTok with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Ismantic/Qwen3-1.7B-Base-ReTok with Transformers:

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

pipe = pipeline("text-generation", model="Ismantic/Qwen3-1.7B-Base-ReTok")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Ismantic/Qwen3-1.7B-Base-ReTok")
model = AutoModelForCausalLM.from_pretrained("Ismantic/Qwen3-1.7B-Base-ReTok", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Ismantic/Qwen3-1.7B-Base-ReTok with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Ismantic/Qwen3-1.7B-Base-ReTok"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Ismantic/Qwen3-1.7B-Base-ReTok",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/Ismantic/Qwen3-1.7B-Base-ReTok

SGLang

How to use Ismantic/Qwen3-1.7B-Base-ReTok 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 "Ismantic/Qwen3-1.7B-Base-ReTok" \
    --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": "Ismantic/Qwen3-1.7B-Base-ReTok",
		"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 "Ismantic/Qwen3-1.7B-Base-ReTok" \
        --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": "Ismantic/Qwen3-1.7B-Base-ReTok",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use Ismantic/Qwen3-1.7B-Base-ReTok with Docker Model Runner:
```
docker model run hf.co/Ismantic/Qwen3-1.7B-Base-ReTok
```

Qwen3-1.7B-Base-ReTok / example_load.py

tf-bao

Update inference code and model card

0c9fe61 verified 3 months ago

Raw History Blame Contribute Delete

1.03 kB

	"""跑通这个发布包 —— 只需要 torch 和 piece_tokenizer。

	pip install torch
	pip install git+https://github.com/Ismantic/PieceTokenizer
	python example_load.py
	"""
	import torch

	from model import Qwen3ForCausalLM
	from tokenizer import PieceTokenizerWrapper

	HERE = "."

	tok = PieceTokenizerWrapper(HERE)
	model = Qwen3ForCausalLM.from_pretrained(
	HERE, device="cuda" if torch.cuda.is_available() else "cpu",
	dtype=torch.bfloat16)

	prompt = "机器翻译的基本任务是"
	ids = tok.encode(prompt, add_special_tokens=False)
	print(f"{prompt!r} -> {len(ids)} tokens")

	# 贪心续写 40 步。没有 KV cache —— 每步重算前缀,短续写够用。
	x = torch.tensor([ids], device=next(model.parameters()).device)
	out = []
	with torch.no_grad():
	for _ in range(40):
	nxt = int(model(x)[0, -1].argmax())
	if nxt == tok.eos_token_id:
	break
	out.append(nxt)
	x = torch.cat([x, torch.tensor([[nxt]], device=x.device)], dim=1)

	print(prompt + tok.decode(out))