Instructions to use OrionStarAI/Orion-14B-LongChat with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use OrionStarAI/Orion-14B-LongChat with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="OrionStarAI/Orion-14B-LongChat", trust_remote_code=True)# Load model directly from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("OrionStarAI/Orion-14B-LongChat", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use OrionStarAI/Orion-14B-LongChat with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "OrionStarAI/Orion-14B-LongChat" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "OrionStarAI/Orion-14B-LongChat", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/OrionStarAI/Orion-14B-LongChat
- SGLang
How to use OrionStarAI/Orion-14B-LongChat 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 "OrionStarAI/Orion-14B-LongChat" \ --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": "OrionStarAI/Orion-14B-LongChat", "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 "OrionStarAI/Orion-14B-LongChat" \ --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": "OrionStarAI/Orion-14B-LongChat", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use OrionStarAI/Orion-14B-LongChat with Docker Model Runner:
docker model run hf.co/OrionStarAI/Orion-14B-LongChat
| from typing import List | |
| from queue import Queue | |
| # build chat input prompt | |
| def build_chat_input(tokenizer, messages: List[dict]): | |
| # chat format: | |
| # single-turn: <s>Human: Hello!\n\nAssistant: </s> | |
| # multi-turn: <s>Human: Hello!\n\nAssistant: </s>Hi!</s>Human: How are you?\n\nAssistant: </s>I'm fine</s> | |
| prompt = "<s>" | |
| for msg in messages: | |
| role = msg["role"] | |
| message = msg["content"] | |
| if message is None : | |
| continue | |
| if role == "user": | |
| prompt += "Human: " + message + "\n\nAssistant: </s>" | |
| if role == "assistant": | |
| prompt += message + "</s>" | |
| input_tokens = tokenizer.encode(prompt) | |
| return input_tokens | |
| class TextIterStreamer: | |
| def __init__(self, tokenizer, skip_prompt=False, skip_special_tokens=False): | |
| self.tokenizer = tokenizer | |
| self.skip_prompt = skip_prompt | |
| self.skip_special_tokens = skip_special_tokens | |
| self.tokens = [] | |
| self.text_queue = Queue() | |
| self.next_tokens_are_prompt = True | |
| def put(self, value): | |
| if self.skip_prompt and self.next_tokens_are_prompt: | |
| self.next_tokens_are_prompt = False | |
| else: | |
| if len(value.shape) > 1: | |
| value = value[0] | |
| self.tokens.extend(value.tolist()) | |
| self.text_queue.put( | |
| self.tokenizer.decode(self.tokens, skip_special_tokens=self.skip_special_tokens)) | |
| def end(self): | |
| self.text_queue.put(None) | |
| def __iter__(self): | |
| return self | |
| def __next__(self): | |
| value = self.text_queue.get() | |
| if value is None: | |
| raise StopIteration() | |
| else: | |
| return value | |