aya expanse 32b
CohereForAIIntroduction
Aya Expanse 32B is a state-of-the-art, multilingual large language model developed by Cohere For AI. It features advanced capabilities across 23 languages and is designed for diverse text generation tasks. The model represents a culmination of extensive research, including data arbitrage, multilingual preference training, safety tuning, and model merging.
Architecture
Aya Expanse 32B is an auto-regressive language model utilizing an optimized transformer architecture. It supports 23 languages, including English, Chinese, Spanish, and Arabic, among others. The model includes a robust post-training process with supervised fine-tuning and preference training to enhance performance.
Training
The model was developed through significant research efforts, focusing on multilingual capabilities. This included using datasets like the Aya Evaluation Suite and m-ArenaHard to evaluate and enhance performance. The model's training involved comparing its capabilities against other significant models such as Gemma 2 27B and Llama 3.1 70B.
Guide: Running Locally
-
Install the Transformers Library: Ensure you have the
transformers
library installed.pip install transformers
-
Load the Model: Use the following Python code to load the Aya Expanse 32B model.
from transformers import AutoTokenizer, AutoModelForCausalLM model_id = "CohereForAI/aya-expanse-32b" tokenizer = AutoTokenizer.from_pretrained(model_id) model = AutoModelForCausalLM.from_pretrained(model_id) messages = [{"role": "user", "content": "Write a letter expressing my love for my mother"}] input_ids = tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True, return_tensors="pt") gen_tokens = model.generate( input_ids, max_new_tokens=100, do_sample=True, temperature=0.3, ) gen_text = tokenizer.decode(gen_tokens[0]) print(gen_text)
-
Cloud GPUs: For optimal performance, consider using cloud GPU services such as AWS EC2, Google Cloud Platform, or Azure.
License
Aya Expanse 32B is released under the CC-BY-NC license, which allows for non-commercial use. Users must also adhere to C4AI's Acceptable Use Policy. For detailed licensing information, visit the license page.