StableVicuna vs. Reka Flash 3

StableVicuna is the first large-scale open-source chatbot that has been developed utilizing reinforced learning from human feedback (RLHF). Building on the Vicuna v0 13b model, it has undergone significant enhancements through further instruction fine-tuning and additional RLHF training. By employing Vicuna as its core model, StableVicuna follows a rigorous three-phase RLHF framework as outlined by researchers Steinnon et al. and Ouyang et al. To achieve its remarkable performance, we engage in further training of the base Vicuna model through supervised fine-tuning (SFT), drawing from a combination of three unique datasets. The first dataset utilized is the OpenAssistant Conversations Dataset (OASST1), which contains 161,443 human-contributed messages organized into 66,497 conversation trees across 35 different languages. The second dataset, known as GPT4All Prompt Generations, includes 437,605 prompts along with responses generated by the GPT-3.5 Turbo model. The final dataset is the Alpaca dataset, featuring 52,000 instructions and examples derived from OpenAI's text-davinci-003 model. This multifaceted training strategy significantly bolsters the chatbot's capability to interact meaningfully across a variety of conversational scenarios, setting a new standard for open-source conversational AI.

Reka Flash 3 stands as a state-of-the-art multimodal AI model, boasting 21 billion parameters and developed by Reka AI, to excel in diverse tasks such as engaging in general conversations, coding, adhering to instructions, and executing various functions. This innovative model skillfully processes and interprets a wide range of inputs, which includes text, images, video, and audio, making it a compact yet versatile solution fit for numerous applications. Constructed from the ground up, Reka Flash 3 was trained on a diverse collection of datasets that include both publicly accessible and synthetic data, undergoing a thorough instruction tuning process with carefully selected high-quality information to refine its performance. The concluding stage of its training leveraged reinforcement learning techniques, specifically the REINFORCE Leave One-Out (RLOO) method, which integrated both model-driven and rule-oriented rewards to enhance its reasoning capabilities significantly. With a remarkable context length of 32,000 tokens, Reka Flash 3 effectively competes against proprietary models such as OpenAI's o1-mini, making it highly suitable for applications that demand low latency or on-device processing. Operating at full precision, the model requires a memory footprint of 39GB (fp16), but this can be optimized down to just 11GB through 4-bit quantization, showcasing its flexibility across various deployment environments. Furthermore, Reka Flash 3's advanced features ensure that it can adapt to a wide array of user requirements, thereby reinforcing its position as a leader in the realm of multimodal AI technology. This advancement not only highlights the progress made in AI but also opens doors to new possibilities for innovation across different sectors.