CognitivessAI/bella-2-8b

Accessible and portable generative AI solutions for developers and businesses.

Website | Demo | GitHub

Description

"Bella-2-8b" by Cognitivess is a text generation model tailored for empathic AI interactions, supporting both English and Romanian languages. The model, built on the transformers architecture, features 8.03 billion parameters , well-suited for a variety of text generation tasks, including question answering, summarization, reasoning, dialogue, sentiment analysis. It employs a floating-point 16 (BF16) tensor type for operations, facilitating speech-to-speech applications. Licensed under Cognitivess AI, Bella-2-8b is available on the Hugging Face platform for wide accessibility.

Intended use

Bella-2-8B is a multilingual chat model designed to support a variety of languages including English, Romanian, Spanish, French, German, and many more, intended for diverse language applications.

Model Developer: Cognitivess AI

Model Dates: Bella-2-8b was trained between May 2024 and June 2024.

Data Freshness: The pretraining data has a cutoff of June 2024. Training will continue beyond the current data cutoff date to incorporate new data as it becomes available.

Model Architecture:

Bella-2-8B model architecture is Transformer-based and trained with a sequence length of 8192 tokens.

Architecture Type: Transformer (auto-regressive language model)

Try this model on bella.cognitivess.com now.

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_path = "CognitivessAI/bella-2-8b"

# Load the tokenizer and model, converting model to half precision
tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(model_path).half().eval()

# Move the model to CUDA if available
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)

# Prompt content: "hi"
messages = [
    {"role": "user", "content": "Who are you?"}
]

input_ids = tokenizer.apply_chat_template(conversation=messages, tokenize=True, add_generation_prompt=True, return_tensors='pt')

# Move input_ids to the same device as the model
input_ids = input_ids.to(device)

# Adjust the generate method to set max_new_tokens
output_ids = model.generate(input_ids, max_new_tokens=50)

response = tokenizer.decode(output_ids[0][input_ids.shape[1]:], skip_special_tokens=True)

# Model response: "I'm Bella, an AI model developed by Cognitivess."
print(response)

Contact: [email protected]