Qwen 2.5
The versatile multilingual powerhouse, excelling in coding and math.
Generic Info
- Publisher: Alibaba Cloud
- Release Date: Sept 2024
- Parameters: 0.5B to 72B
- Context Window: 128k tokens
- License: Apache 2.0 (mostly)
- Key Capabilities: Multilingual, Coding, Math, Instruction Following
Qwen 2.5 is a comprehensive suite of models that rival GPT-4 in many benchmarks, particularly in coding and mathematics. Its strong performance across diverse languages makes it a top choice for global applications.
Hello World Guide
Run Qwen 2.5 locally using Hugging Face transformers.
Python
from transformers import AutoModelForCausalLM, AutoTokenizer
model_name = "Qwen/Qwen2.5-7B-Instruct"
model = AutoModelForCausalLM.from_pretrained(
model_name,
torch_dtype="auto",
device_map="auto"
)
tokenizer = AutoTokenizer.from_pretrained(model_name)
prompt = "Write a quick sort algorithm in Python."
messages = [
{"role": "system", "content": "You are a helpful coding assistant."},
{"role": "user", "content": prompt}
]
text = tokenizer.apply_chat_template(
messages,
tokenize=False,
add_generation_prompt=True
)
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
generated_ids = model.generate(
**model_inputs,
max_new_tokens=512
)
generated_ids = [
output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
]
response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
print(response)
Industry Usage
Coding Assistants
Powering IDE plugins and internal developer tools due to its exceptional grasp of programming languages.
Multilingual Content
Used by media companies to generate and translate content across Asian and Western languages with high fidelity.
Mobile Applications
Smaller variants (0.5B, 1.5B, 3B) are deployed on mobile devices for offline AI features.