Instructions to use ZeroLoss-Lab/Innospark-72b-safety with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ZeroLoss-Lab/Innospark-72b-safety with Transformers:

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

pipe = pipeline("text-generation", model="ZeroLoss-Lab/Innospark-72b-safety")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("ZeroLoss-Lab/Innospark-72b-safety")
model = AutoModelForCausalLM.from_pretrained("ZeroLoss-Lab/Innospark-72b-safety", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use ZeroLoss-Lab/Innospark-72b-safety with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ZeroLoss-Lab/Innospark-72b-safety"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ZeroLoss-Lab/Innospark-72b-safety",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ZeroLoss-Lab/Innospark-72b-safety

SGLang

How to use ZeroLoss-Lab/Innospark-72b-safety 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 "ZeroLoss-Lab/Innospark-72b-safety" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ZeroLoss-Lab/Innospark-72b-safety",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "ZeroLoss-Lab/Innospark-72b-safety" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ZeroLoss-Lab/Innospark-72b-safety",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ZeroLoss-Lab/Innospark-72b-safety with Docker Model Runner:
```
docker model run hf.co/ZeroLoss-Lab/Innospark-72b-safety
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Innospark-72b-safety

所属项目

本模型是 InnoSpark-Safety 教育场景大语言模型安全框架中的安全对齐底座模型。该框架围绕用户输入、模型生成与服务化调用三个环节，提供前置拦截、安全对齐、后置审核与 OpenAI 兼容 API 代理能力。

模型概述

本仓库包含一个基于 Qwen/Qwen2.5-72B、通过强化学习（RL）训练得到的安全对齐 72B 参数因果语言模型检查点。

属性	值
架构	Qwen2ForCausalLM
基座模型	Qwen/Qwen2.5-72B
隐藏层维度	8192
层数	80
注意力头（GQA）	64 query / 8 key-value
FFN 中间维度	29568
词表大小	152064
上下文长度	32768
滑动窗口	131072
权重类型	bfloat16
格式	Safetensors（31 个分片，约 145 GB）

文件清单

config.json、generation_config.json、tokenizer_config.json
tokenizer.json、vocab.json、merges.txt、added_tokens.json、special_tokens_map.json
model.safetensors.index.json
model-00001-of-00031.safetensors ... model-00031-of-00031.safetensors
.gitattributes

使用方法

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "ZeroLoss-Lab/Innospark-72b-safety"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype="auto",
    device_map="auto",
)

messages = [
    {"role": "system", "content": "You are a helpful assistant."},
    {"role": "user", "content": "Hello, what can you do?"},
]
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer([text], return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=512)
response = tokenizer.batch_decode(outputs, skip_special_tokens=True)[0]
print(response)

生成配置

Temperature: 0.7
Top-p: 0.8
Top-k: 20
Repetition penalty: 1.05
EOS token IDs: 151645, 151643

许可证

本模型采用 Apache-2.0 许可证发布。

Downloads last month: -

Safetensors

Model size

73B params

Tensor type

BF16

Model tree for ZeroLoss-Lab/Innospark-72b-safety

Base model

Qwen/Qwen2.5-72B

Finetuned

(68)

this model

Quantizations

1 model

ZeroLoss-Lab
/

Innospark-72b-safety