Instructions to use QuantTrio/DeepSeek-V3.2-Speciale-AWQ with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use QuantTrio/DeepSeek-V3.2-Speciale-AWQ with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="QuantTrio/DeepSeek-V3.2-Speciale-AWQ") messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("QuantTrio/DeepSeek-V3.2-Speciale-AWQ") model = AutoModelForCausalLM.from_pretrained("QuantTrio/DeepSeek-V3.2-Speciale-AWQ", 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 QuantTrio/DeepSeek-V3.2-Speciale-AWQ with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "QuantTrio/DeepSeek-V3.2-Speciale-AWQ" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "QuantTrio/DeepSeek-V3.2-Speciale-AWQ", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/QuantTrio/DeepSeek-V3.2-Speciale-AWQ
- SGLang
How to use QuantTrio/DeepSeek-V3.2-Speciale-AWQ 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 "QuantTrio/DeepSeek-V3.2-Speciale-AWQ" \ --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": "QuantTrio/DeepSeek-V3.2-Speciale-AWQ", "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 "QuantTrio/DeepSeek-V3.2-Speciale-AWQ" \ --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": "QuantTrio/DeepSeek-V3.2-Speciale-AWQ", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use QuantTrio/DeepSeek-V3.2-Speciale-AWQ with Docker Model Runner:
docker model run hf.co/QuantTrio/DeepSeek-V3.2-Speciale-AWQ
File size: 2,120 Bytes
19cdb6f | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 | import json
import copy
from encoding_dsv32 import encode_messages, parse_message_from_completion_text
with open("test_input.json", "r") as f:
test_dict = json.load(f)
messages = test_dict["messages"]
messages[0]["tools"] = test_dict["tools"]
with open("test_output.txt", "r") as f:
gold_prompt = f.read().strip()
print(messages)
print("=" * 60)
encode_config = dict(thinking_mode="thinking", drop_thinking=True, add_default_bos_token=True)
prompt = encode_messages(messages, **encode_config)
print(prompt)
assert prompt == gold_prompt
print("=" * 60)
tool_call_message = messages[4]
tool_call_prompt = encode_messages([tool_call_message], context=messages[:4], **encode_config)
tool_call_message_wo_id = copy.deepcopy(tool_call_message)
for tool_call in tool_call_message_wo_id["tool_calls"]:
tool_call.pop("id")
parsed_tool_call_message = parse_message_from_completion_text(tool_call_prompt, thinking_mode="thinking")
parsed_tool_call_message.pop("content")
assert tool_call_message_wo_id == parsed_tool_call_message
thinking_message = messages[-6]
thinking_prompt = encode_messages([thinking_message], context=messages[:-6], **encode_config)
parsed_thinking_message = parse_message_from_completion_text(thinking_prompt, thinking_mode="thinking")
parsed_thinking_message.pop("tool_calls")
assert thinking_message == parsed_thinking_message
with open("test_input_search_wo_date.json", "r") as f:
search_messages = json.load(f)["messages"]
with open("test_output_search_wo_date.txt", "r") as f:
search_gold_prompt = f.read().strip()
search_prompt = encode_messages(search_messages, **encode_config)
assert search_prompt == search_gold_prompt
with open("test_input_search_w_date.json", "r") as f:
search_messages_w_date = json.load(f)["messages"]
with open("test_output_search_w_date.txt", "r") as f:
search_gold_prompt_w_date = f.read().strip()
search_prompt_w_date = encode_messages(search_messages_w_date, **encode_config)
with open("test_output_search_w_date_2.txt", "w") as f:
f.write(search_prompt_w_date)
assert search_prompt_w_date == search_gold_prompt_w_date |