Skip to main content
Back

Meta's New Code Llama 70B Outperforms GPT-4 Open Source

GET YOUR GPU

January 30, 2024 4 min read

Hey 

Welcome to this week's edition of AlphaSignal. 


Whether you are a researcher, engineer, developer, or data scientist, our summaries ensure you're always up-to-date with the latest breakthroughs in AI. 


Let's get into it!


Lior

IN TODAY'S EMAIL

  • Top Announcements

  • Meta's new model outperforms GPT-4

  • Eagle 7B a new RNN-based LLM model

Read Time: 4 min 10 sec

RELEASES AND ANNOUNCEMENTS

OpenAI Introduces New Embedding Model and API Upgrades

---------------------
OpenAI released new embedding models (text-embedding-3-small/large), updated GPT-4/3.5 Turbo, reduced prices, and enhanced API management tools. text-embedding-3-small/large showed improved MIRACL (44.0%/54.9%) and MTEB (62.3%/64.6%) scores, with scalable dimensions and cost-efficiency.



Google Releases New Bard Gemini Model that is on par with GPT-4 in Human Evaluation

---------------------
Google released Bard Gemini Pro, matching GPT-4 in human evaluations. It ranked second on Chatbot Arena, surpassing GPT-4 03/14 and 06/13, but trailing GPT-4 Turbo. The model's performance is notable, though based on 3,000 user ratings, indicating potential for shifts with more data.



Adept introduces Fuyu-Heavy: A New Multimodal Model Designed Specifically for Digital Agents

---------------------
Adept introduced Fuyu-Heavy, the third-most-capable multimodal model, excelling in UI understanding and outperforming Gemini Pro in the MMMU benchmark. Despite its smaller size, it matches text-only benchmarks of larger models and integrates text-image processing, indicating high efficiency and potential for diverse applications.



Google Releases Lumiere, A New AI Model that Creates Realistic Videos

---------------------

Google's Lumiere, a space-time diffusion model, generates realistic videos using STUNet architecture for coherent motion, outperforming existing T2V models by synthesizing 80-frame, 5-second videos at 16 fps, trained on 30 million videos, and addressing temporal consistency challenges.



Hugging Face and Google Partner for Open AI Collaboration

---------------------

Hugging Face and Google Cloud partner to integrate open AI models and cloud services, enabling AI development through Vertex AI, GKE, and enhanced hardware (TPU v5e, A3 VMs, C3 VMs) for training, tuning, and deploying AI applications.


Get On-demand Cloud GPUs: Train Heavy Models in Seconds

Accelerate your training and inference workloads with the power of NVIDIA’s H100 Tensor Core, deploy your instances in seconds with Latitude.sh


High-performance networking and enterprise-grade hardware designed for Machine Learning engineers. Our super intuitive dashboard and pre-installed AI software gets you from zero to training with just a few clicks.


Take full advantage of the H100 Tensor Core:

  • Up to 4X higher AI Training on GPT-3
  • Up to 7X higher performance for HPC applications
  • Up to 30X higher AI inference performance on the largest models (530 billion parameters)

Get instant access to on-demand plans with no upfront commitment and deploy clusters according to your specifics needs: 8xH100 and 1xH100 options available now.

Get your gpu

partner with us

NEWS

Meta releases ‘Code Llama 70B’ Outperforming GPT-4 on HumanEval

What's New?

Meta has unveiled Code Llama 70B, a leap in open-source code generation models, surpassing GPT-4 with a HumanEval score of 67.8. Building on Llama 2's architecture, Code Llama 70B, a large language model (LLM), specializes in code-related tasks, offering enhanced code generation and natural language understanding of code.


Code Llama 70B is trained on 1TB tokens of code and code-related data, setting a new standard in the field. It's designed to generate code from text prompts, supporting popular programming languages like Python, C++, and Java, and is especially proficient in Python, owing to additional fine-tuning on 100B tokens of Python code.


This model offers three versions: a foundational code model, a Python-specific model, and an Instruct model fine-tuned for natural language instructions. The Instruct variant shows improved performance in generating contextually accurate and safe code outputs, making it particularly useful for code generation and debugging tasks.


Code Llama models handle up to 100,000 tokens of context, improving code generation for larger projects. They also exhibit versatility in serving different latency requirements, with the smaller 7B and 13B models offering faster, real-time code completion.

Read more

NEWS

RWKV Launches Eagle 7B; An RNN-based LLM that Outperforms Transformers

What's New?


Eagle-7B, a new player in the large language model (LLM) space, is shaking up the field with its innovative RWKV-v5 architecture. This model diverges from traditional transformer structures, opting instead for an RNN-based design that significantly reduces inference costs by a factor of 10 to 100 times. 


Notably, Eagle-7B excels in multi-lingual benchmarks, outperforming all models in its class and nearing the prowess of heavyweights like Falcon, LLaMA2, and Mistral in English evaluations.


Developed on the backbone of 1.1 trillion tokens across over 100 languages, Eagle-7B's strength lies in its broad linguistic reach and energy-efficient design, earning it the title of the "world's greenest 7B model." Its open-source nature under the Apache 2.0 license further democratizes access, allowing both individual and commercial use without restrictions.


Eagle-7B is not just about raw power; it's a foundation model that requires minimal instruct tuning, making it adaptable to a wide array of use cases. Its performance in both English and multi-lingual benchmarks indicates a significant leap from its predecessor, RWKV-v4, showcasing the potential of linear transformers and RNN-based models in challenging the dominance of traditional transformer architectures. 


With plans for further training and fine-tuning, Eagle-7B is poised to redefine performance standards in the LLM landscape.


Download the model

Read more

Not Great   Good    Amazing

Thank You.

Want to promote your company, product, job, or event to 150,000+ AI researchers and engineers? You can reach out here.