DeciLM 6B: The New Frontier in Efficient Large Language Models

October 3, 2023


In recent years, the field of Natural Language Processing (NLP) has witnessed a paradigm shift with the advent of Large Language Models (LLMs). These models, characterized by their massive number of parameters and deep architectures, have set new benchmarks in a wide array of NLP tasks, from machine translation to question-answering. However, as these models grow in size and complexity, they bring along their own set of challenges.

The computational demands for training and deploying LLMs have skyrocketed, leading to increased costs and energy consumption. These challenges have not only financial implications but also environmental ones, given the carbon footprint associated with running these models. Furthermore, the latency in real-world applications can be a bottleneck, affecting the user experience and limiting the scalability of services built on top of these models. Therefore, there is an urgent need to find a balance between computational efficiency and model performance.

DeciLM 6B is a groundbreaking model that promises to redefine the landscape of LLMs. Developed by Deci, this model boasts 5.7 billion parameters yet delivers a throughput that's 15 times higher than its contemporaries like Llama 2 7B. What sets DeciLM 6B apart is its unique architecture, featuring a decoder-only transformer with variable Grouped-Query Attention (GQA), all fine-tuned to offer unparalleled efficiency and performance.

In this blog, we will discuss the intricacies of DeciLM 6B, explore its technological underpinnings, and discuss its potential impact on the future.

What Makes DeciLM 6B Stand Out?

Any LLM is surrounded by competition, each boasting billions of parameters and impressive performance metrics. Amidst this competitive backdrop, DeciLM 6B emerges as a standout, not just for its size but for its unique blend of efficiency, performance, and architectural innovation. Let's delve into the key features that set DeciLM 6B apart.

At first glance, 5.7 billion parameters might seem like just another big number in the world of LLMs. However, what makes this figure truly remarkable is how DeciLM 6B leverages these parameters for optimized performance. The model manages to maintain a high level of quality in its outputs, comparable to models with even larger parameter counts, such as Llama 2 7B. This efficiency in parameter utilization makes DeciLM 6B a cost-effective and computationally efficient option for a variety of NLP tasks.

One of the most striking features of DeciLM 6B is its throughput, which is 15 times higher than that of Llama 2 7B. This is a game-changer for real-world applications where latency and speed are critical. Whether it's customer service chatbots, real-time translation services, or content generation, the high throughput of DeciLM 6B ensures that applications can handle larger volumes of data more efficiently, thereby improving user experience and operational scalability. 

While Deci's team claims that DeciLM 6B offers a throughput that's 15 times higher than Llama 2 7B, it's important to note that these figures come from the developers themselves. Independent verification by external researchers or organizations is crucial for substantiating these claims. Such third-party evaluations can provide a more comprehensive understanding of the model's performance and its practical implications.

The architecture of DeciLM 6B is another feather in its cap. Unlike many LLMs that use a standard transformer architecture, DeciLM 6B employs a decoder-only transformer. This design choice is pivotal for its efficiency, but what truly sets it apart is its implementation of variable Grouped-Query Attention (GQA). This novel attention mechanism allows the model to dynamically adjust its attention patterns, optimizing for both computational efficiency and the quality of the generated text. This is made possible through Deci's proprietary Neural Architecture Search engine, AutoNAC, which fine-tunes the architecture for peak performance.

The Technology Behind DeciLM 6B

DeciLM 6B is not just another large language model; it's a technological marvel built on cutting-edge innovations in neural architecture and machine learning. In this section, we'll explore the key technologies that power DeciLM 6B, making it one of the most efficient and high-performing models in its class.

AutoNAC: Deci's Neural Architecture Search Engine

One of the standout features of DeciLM 6B is its architecture, which was generated using Deci's proprietary Neural Architecture Search (NAS) engine, AutoNAC. Traditional NAS methods are computationally intensive and can take a significant amount of time and resources. AutoNAC, however, automates this process in a more compute-efficient manner, enabling the selection of optimal architectural elements for each layer of the model. This is pivotal for achieving the high throughput and efficiency that DeciLM 6B is known for.

Variable Grouped-Query Attention (GQA): A Closer Look

The attention mechanism is a cornerstone of any transformer-based model, and DeciLM 6B is no exception. What sets it apart is its unique implementation of variable Grouped-Query Attention (GQA). Unlike standard models that maintain consistent attention groups across layers, DeciLM 6B introduces variability, allowing for a more dynamic and efficient allocation of computational resources. This results in a model that can maintain high-quality outputs while being more computationally efficient.

Training on the SlimPajamas Dataset

The efficacy of any machine learning model is heavily dependent on the quality of the data it's trained on. DeciLM 6B was trained on a subset of the SlimPajamas dataset, an extensive, deduplicated, multi-corpora open-source dataset. This dataset has been carefully curated to include a diverse range of text, enabling DeciLM 6B to generalize well across various NLP tasks. The model also underwent LoRA fine-tuning, further enhancing its performance and making it suitable for instruction-following use-cases.

Performance Metrics

In the realm of Large Language Models, performance metrics are the yardstick by which we measure a model's capabilities and limitations. DeciLM 6B, despite having fewer parameters than some of its competitors in the 7-billion parameter class, has shown remarkable performance across various benchmarks. Let's delve into how DeciLM 6B stacks up against other models and examine its memory efficiency and throughput.

Benchmarking Against Other Models in the 7-Billion Parameter Class

DeciLM 6B has been benchmarked against leading models like Llama 2 7B, Falcon 7B, and MPT-Instruct. Impressively, DeciLM 6B-Instruct, the fine-tuned version of DeciLM 6B, clinched the third spot in performance metrics, trailing Llama 2 7B by just under a percentage point. This is a significant achievement, considering DeciLM 6B operates with fewer parameters, yet manages to deliver comparable, if not superior, performance in various NLP tasks.

Memory Efficiency and Throughput Comparisons

One of the most compelling aspects of DeciLM 6B is its memory efficiency. When tested on an NVIDIA A10G GPU, DeciLM 6B was able to handle a batch size of 64, compared to Llama 2 7B's maximum batch size of 8. This larger batch size allows for better utilization of GPU resources, making DeciLM 6B a more cost-effective choice for deployment.

In terms of throughput, DeciLM 6B has set new standards. Its throughput, measured in tokens per second, is 4.8 times that of Llama 2 7B when tested under similar conditions. This high throughput makes DeciLM 6B ideal for applications requiring real-time responses and high data volumes.

Infery-LLM: Turbocharging DeciLM 6B

While DeciLM 6B already stands as a paragon of efficiency and performance, its capabilities are further amplified when coupled with Infery-LLM, Deci's specialized inference SDK. In this section, we'll explore how Infery-LLM acts as a turbocharger for DeciLM 6B, enhancing its throughput, reducing costs, and contributing to a more sustainable computational environment.

Infery-LLM is not just another inference tool; it's a game-changer in the world of Large Language Models. Built on advanced engineering techniques like selective quantization and hybrid compilation, Infery-LLM is designed to accelerate the computational processing of LLMs. It incorporates proprietary optimized kernels, fast beam search algorithms, and other features that make it an instrumental tool for boosting the performance of models like DeciLM 6B.

  • Throughput Boost: When integrated with DeciLM 6B, Infery-LLM delivers a throughput that is a staggering 15 times higher than that of Llama 2 7B when tested on NVIDIA's A10G GPU. This is a monumental leap, especially for applications where real-time responses are crucial.
  • Cost Implications: The efficiency gains from using Infery-LLM are not just theoretical; they have real-world cost benefits. By enabling workloads to migrate from more expensive GPUs like the A100 to the more affordable A10G, Infery-LLM significantly reduces the cost of running large language models.
  • Environmental Impact: The computational efficiency of Infery-LLM also has a positive environmental impact. By reducing the computational resources needed for inference, Infery-LLM contributes to a reduction in carbon footprint, estimated at 516 kg CO2 per model instance yearly when running on the A10G GPU.

Practical Applications

DeciLM 6B isn't just a theoretical marvel; it has practical implications that extend across both commercial and research domains. In this section, we'll explore some of the key use-cases where DeciLM 6B can make a significant impact.

  • Customer Service Chatbots: With its high throughput and low latency, DeciLM 6B is ideal for powering customer service chatbots that require real-time responses.
  • Content Generation: Whether it's automated journalism or scriptwriting, DeciLM 6B's high-quality text generation capabilities make it a valuable tool in content creation.
  • Machine Translation: The model's efficiency and accuracy make it well-suited for real-time translation services, a boon for both commercial applications and academic research.
  • Sentiment Analysis: DeciLM 6B can be fine-tuned to perform sentiment analysis, providing valuable insights for market research and social listening tools.

DeciLM 6B's architecture and large parameter count make it highly adaptable for fine-tuning. Researchers and developers can leverage transfer learning techniques to adapt the model for specific tasks, from text summarization to medical diagnosis based on textual data.

Getting Started with DeciLM 6B

Below are some resources and code snippets to help run the program.

# pip install -q transformers

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

checkpoint = "Deci/DeciLM-6b"
device = "cuda" # for GPU usage or "cpu" for CPU usage

tokenizer = AutoTokenizer.from_pretrained(checkpoint)
model = AutoModelForCausalLM.from_pretrained(checkpoint, torch_dtype=torch.bfloat16, trust_remote_code=True).to(device)

# Sample code for text generation
inputs = tokenizer.encode("In a shocking finding, scientists discovered a herd of unicorns living in", return_tensors="pt").to(device)
outputs = model.generate(inputs, max_new_tokens=100, do_sample=True, top_p=0.95)

DeciLM 6B is available under the Llama 2 Community License Agreement, with an extension from Deci regarding hosting service providers. This permissive license makes it accessible for both commercial and academic use.


In a landscape teeming with Large Language Models, DeciLM 6B emerges not just as another contender, but as a game-changer. With its unique blend of computational efficiency, high throughput, and architectural innovation, DeciLM 6B sets new benchmarks for what is achievable in the realm of natural language processing.

The model's 5.7 billion parameters are optimized for peak performance, delivering a throughput that's 15 times higher than some of its closest competitors. Its unique architecture, featuring a decoder-only transformer with variable Grouped-Query Attention, is a testament to the kind of innovation that pushes the boundaries of what LLMs can do. When coupled with Infery-LLM, DeciLM 6B's capabilities are further amplified, making it an ideal choice for a wide range of commercial and research applications.

Future Prospects and Developments to Look Out For

As impressive as DeciLM 6B is, it's just the tip of the iceberg. The model's architecture lends itself well to further fine-tuning and adaptation, opening the door for even more specialized and efficient models in the future. With ongoing research and the potential for community contributions, we can expect to see even more groundbreaking developments in the coming months and years.

In a world where data is abundant but attention is scarce, DeciLM 6B stands out as a beacon of efficiency and performance. It not only addresses the current challenges faced by LLMs but also lays the groundwork for the future of efficient, high-performing AI models.

Latest Blogs
This is a decorative image for: A Complete Guide To Customer Acquisition For Startups
October 18, 2022

A Complete Guide To Customer Acquisition For Startups

Any business is enlivened by its customers. Therefore, a strategy to constantly bring in new clients is an ongoing requirement. In this regard, having a proper customer acquisition strategy can be of great importance.

So, if you are just starting your business, or planning to expand it, read on to learn more about this concept.

The problem with customer acquisition

As an organization, when working in a diverse and competitive market like India, you need to have a well-defined customer acquisition strategy to attain success. However, this is where most startups struggle. Now, you may have a great product or service, but if you are not in the right place targeting the right demographic, you are not likely to get the results you want.

To resolve this, typically, companies invest, but if that is not channelized properly, it will be futile.

So, the best way out of this dilemma is to have a clear customer acquisition strategy in place.

How can you create the ideal customer acquisition strategy for your business?

  • Define what your goals are

You need to define your goals so that you can meet the revenue expectations you have for the current fiscal year. You need to find a value for the metrics –

  • MRR – Monthly recurring revenue, which tells you all the income that can be generated from all your income channels.
  • CLV – Customer lifetime value tells you how much a customer is willing to spend on your business during your mutual relationship duration.  
  • CAC – Customer acquisition costs, which tells how much your organization needs to spend to acquire customers constantly.
  • Churn rate – It tells you the rate at which customers stop doing business.

All these metrics tell you how well you will be able to grow your business and revenue.

  • Identify your ideal customers

You need to understand who your current customers are and who your target customers are. Once you are aware of your customer base, you can focus your energies in that direction and get the maximum sale of your products or services. You can also understand what your customers require through various analytics and markers and address them to leverage your products/services towards them.

  • Choose your channels for customer acquisition

How will you acquire customers who will eventually tell at what scale and at what rate you need to expand your business? You could market and sell your products on social media channels like Instagram, Facebook and YouTube, or invest in paid marketing like Google Ads. You need to develop a unique strategy for each of these channels. 

  • Communicate with your customers

If you know exactly what your customers have in mind, then you will be able to develop your customer strategy with a clear perspective in mind. You can do it through surveys or customer opinion forms, email contact forms, blog posts and social media posts. After that, you just need to measure the analytics, clearly understand the insights, and improve your strategy accordingly.

Combining these strategies with your long-term business plan will bring results. However, there will be challenges on the way, where you need to adapt as per the requirements to make the most of it. At the same time, introducing new technologies like AI and ML can also solve such issues easily. To learn more about the use of AI and ML and how they are transforming businesses, keep referring to the blog section of E2E Networks.

Reference Links

This is a decorative image for: Constructing 3D objects through Deep Learning
October 18, 2022

Image-based 3D Object Reconstruction State-of-the-Art and trends in the Deep Learning Era

3D reconstruction is one of the most complex issues of deep learning systems. There have been multiple types of research in this field, and almost everything has been tried on it — computer vision, computer graphics and machine learning, but to no avail. However, that has resulted in CNN or convolutional neural networks foraying into this field, which has yielded some success.

The Main Objective of the 3D Object Reconstruction

Developing this deep learning technology aims to infer the shape of 3D objects from 2D images. So, to conduct the experiment, you need the following:

  • Highly calibrated cameras that take a photograph of the image from various angles.
  • Large training datasets can predict the geometry of the object whose 3D image reconstruction needs to be done. These datasets can be collected from a database of images, or they can be collected and sampled from a video.

By using the apparatus and datasets, you will be able to proceed with the 3D reconstruction from 2D datasets.

State-of-the-art Technology Used by the Datasets for the Reconstruction of 3D Objects

The technology used for this purpose needs to stick to the following parameters:

  • Input

Training with the help of one or multiple RGB images, where the segmentation of the 3D ground truth needs to be done. It could be one image, multiple images or even a video stream.

The testing will also be done on the same parameters, which will also help to create a uniform, cluttered background, or both.

  • Output

The volumetric output will be done in both high and low resolution, and the surface output will be generated through parameterisation, template deformation and point cloud. Moreover, the direct and intermediate outputs will be calculated this way.

  • Network architecture used

The architecture used in training is 3D-VAE-GAN, which has an encoder and a decoder, with TL-Net and conditional GAN. At the same time, the testing architecture is 3D-VAE, which has an encoder and a decoder.

  • Training used

The degree of supervision used in 2D vs 3D supervision, weak supervision along with loss functions have to be included in this system. The training procedure is adversarial training with joint 2D and 3D embeddings. Also, the network architecture is extremely important for the speed and processing quality of the output images.

  • Practical applications and use cases

Volumetric representations and surface representations can do the reconstruction. Powerful computer systems need to be used for reconstruction.

Given below are some of the places where 3D Object Reconstruction Deep Learning Systems are used:

  • 3D reconstruction technology can be used in the Police Department for drawing the faces of criminals whose images have been procured from a crime site where their faces are not completely revealed.
  • It can be used for re-modelling ruins at ancient architectural sites. The rubble or the debris stubs of structures can be used to recreate the entire building structure and get an idea of how it looked in the past.
  • They can be used in plastic surgery where the organs, face, limbs or any other portion of the body has been damaged and needs to be rebuilt.
  • It can be used in airport security, where concealed shapes can be used for guessing whether a person is armed or is carrying explosives or not.
  • It can also help in completing DNA sequences.

So, if you are planning to implement this technology, then you can rent the required infrastructure from E2E Networks and avoid investing in it. And if you plan to learn more about such topics, then keep a tab on the blog section of the website

Reference Links

This is a decorative image for: Comprehensive Guide to Deep Q-Learning for Data Science Enthusiasts
October 18, 2022

A Comprehensive Guide To Deep Q-Learning For Data Science Enthusiasts

For all data science enthusiasts who would love to dig deep, we have composed a write-up about Q-Learning specifically for you all. Deep Q-Learning and Reinforcement learning (RL) are extremely popular these days. These two data science methodologies use Python libraries like TensorFlow 2 and openAI’s Gym environment.

So, read on to know more.

What is Deep Q-Learning?

Deep Q-Learning utilizes the principles of Q-learning, but instead of using the Q-table, it uses the neural network. The algorithm of deep Q-Learning uses the states as input and the optimal Q-value of every action possible as the output. The agent gathers and stores all the previous experiences in the memory of the trained tuple in the following order:

State> Next state> Action> Reward

The neural network training stability increases using a random batch of previous data by using the experience replay. Experience replay also means the previous experiences stocking, and the target network uses it for training and calculation of the Q-network and the predicted Q-Value. This neural network uses openAI Gym, which is provided by taxi-v3 environments.

Now, any understanding of Deep Q-Learning   is incomplete without talking about Reinforcement Learning.

What is Reinforcement Learning?

Reinforcement is a subsection of ML. This part of ML is related to the action in which an environmental agent participates in a reward-based system and uses Reinforcement Learning to maximize the rewards. Reinforcement Learning is a different technique from unsupervised learning or supervised learning because it does not require a supervised input/output pair. The number of corrections is also less, so it is a highly efficient technique.

Now, the understanding of reinforcement learning is incomplete without knowing about Markov Decision Process (MDP). MDP is involved with each state that has been presented in the results of the environment, derived from the state previously there. The information which composes both states is gathered and transferred to the decision process. The task of the chosen agent is to maximize the awards. The MDP optimizes the actions and helps construct the optimal policy.

For developing the MDP, you need to follow the Q-Learning Algorithm, which is an extremely important part of data science and machine learning.

What is Q-Learning Algorithm?

The process of Q-Learning is important for understanding the data from scratch. It involves defining the parameters, choosing the actions from the current state and also choosing the actions from the previous state and then developing a Q-table for maximizing the results or output rewards.

The 4 steps that are involved in Q-Learning:

  1. Initializing parameters – The RL (reinforcement learning) model learns the set of actions that the agent requires in the state, environment and time.
  2. Identifying current state – The model stores the prior records for optimal action definition for maximizing the results. For acting in the present state, the state needs to be identified and perform an action combination for it.
  3. Choosing the optimal action set and gaining the relevant experience – A Q-table is generated from the data with a set of specific states and actions, and the weight of this data is calculated for updating the Q-Table to the following step.
  4. Updating Q-table rewards and next state determination – After the relevant experience is gained and agents start getting environmental records. The reward amplitude helps to present the subsequent step.  

In case the Q-table size is huge, then the generation of the model is a time-consuming process. This situation requires Deep Q-learning.

Hopefully, this write-up has provided an outline of Deep Q-Learning and its related concepts. If you wish to learn more about such topics, then keep a tab on the blog section of the E2E Networks website.

Reference Links

This is a decorative image for: GAUDI: A Neural Architect for Immersive 3D Scene Generation
October 13, 2022

GAUDI: A Neural Architect for Immersive 3D Scene Generation

The evolution of artificial intelligence in the past decade has been staggering, and now the focus is shifting towards AI and ML systems to understand and generate 3D spaces. As a result, there has been extensive research on manipulating 3D generative models. In this regard, Apple’s AI and ML scientists have developed GAUDI, a method specifically for this job.

An introduction to GAUDI

The GAUDI 3D immersive technique founders named it after the famous architect Antoni Gaudi. This AI model takes the help of a camera pose decoder, which enables it to guess the possible camera angles of a scene. Hence, the decoder then makes it possible to predict the 3D canvas from almost every angle.

What does GAUDI do?

GAUDI can perform multiple functions –

  • The extensions of these generative models have a tremendous effect on ML and computer vision. Pragmatically, such models are highly useful. They are applied in model-based reinforcement learning and planning world models, SLAM is s, or 3D content creation.
  • Generative modelling for 3D objects has been used for generating scenes using graf, pigan, and gsn, which incorporate a GAN (Generative Adversarial Network). The generator codes radiance fields exclusively. Using the 3D space in the scene along with the camera pose generates the 3D image from that point. This point has a density scalar and RGB value for that specific point in 3D space. This can be done from a 2D camera view. It does this by imposing 3D datasets on those 2D shots. It isolates various objects and scenes and combines them to render a new scene altogether.
  • GAUDI also removes GANs pathologies like mode collapse and improved GAN.
  • GAUDI also uses this to train data on a canonical coordinate system. You can compare it by looking at the trajectory of the scenes.

How is GAUDI applied to the content?

The steps of application for GAUDI have been given below:

  • Each trajectory is created, which consists of a sequence of posed images (These images are from a 3D scene) encoded into a latent representation. This representation which has a radiance field or what we refer to as the 3D scene and the camera path is created in a disentangled way. The results are interpreted as free parameters. The problem is optimized by and formulation of a reconstruction objective.
  • This simple training process is then scaled to trajectories, thousands of them creating a large number of views. The model samples the radiance fields totally from the previous distribution that the model has learned.
  • The scenes are thus synthesized by interpolation within the hidden space.
  • The scaling of 3D scenes generates many scenes that contain thousands of images. During training, there is no issue related to canonical orientation or mode collapse.
  • A novel de-noising optimization technique is used to find hidden representations that collaborate in modelling the camera poses and the radiance field to create multiple datasets with state-of-the-art performance in generating 3D scenes by building a setup that uses images and text.

To conclude, GAUDI has more capabilities and can also be used for sampling various images and video datasets. Furthermore, this will make a foray into AR (augmented reality) and VR (virtual reality). With GAUDI in hand, the sky is only the limit in the field of media creation. So, if you enjoy reading about the latest development in the field of AI and ML, then keep a tab on the blog section of the E2E Networks website.

Reference Links

Build on the most powerful infrastructure cloud

A vector illustration of a tech city using latest cloud technologies & infrastructure