General Introduction
Cerebras is a company dedicated to advancing the field of artificial intelligence and high-performance computing. Its core products include the world's fastest AI inference platform and high-performance computing gas pedal.The Cerebras platform is capable of training a wide range of models, from multi-lingual macromodels to medical chatbots, helping customers train their own base models or fine-tune open-source models.Cerebras's technology offers significant advantages in terms of speed and cost, and is capable of supporting hundreds of concurrent users, delivering high-throughput and low-cost AI inference services.
Currently faster than Groq, Llama 3.1 8B can generate up to 1800 Token per second for Sudoku, and 70B can reach 450 Token.
Function List
- AI Reasoning: provides the world's fastest AI reasoning service, supporting a variety of large models.
- High Performance Computing: With 900,000 cores and 44GB of on-chip memory, redefining the performance of high performance computing systems.
- Model Training: Help customers train multiple base models or fine-tune open source models.
- Open source support: most of the work is open source and supports community sharing and collaboration.
Using Help
Installation and use
The Cerebras platform does not require a complicated installation process, users can simply register and log in to their account via the official website to start using its AI reasoning and HPC services. Here are the detailed steps to use it:
- register an accountVisit the Cerebras website, click the "Register" button in the upper right corner, and fill in the relevant information to complete the registration.
- Log in to your account: Log in to the Cerebras platform using your registered email and password.
- Select Service: On the platform's home page, select the service module you need to use, such as AI reasoning or high-performance computing.
- Upload data: Upload data files to be processed according to the prompts, supporting multiple data formats.
- Configuration parametersThe platform provides a wide range of preset options to choose from: configure the parameters for model training or inference according to requirements.
- Commencement of missionClick the "Start" button and the platform will automatically process the task and display the progress in real time.
- View Results: After the task is completed, users can view detailed processing results and analysis reports on the Results page.
Functional operation flow
- AI reasoning::
- Select the "AI Reasoning" module.
- Upload a data file that requires inference.
- Choose either a pre-trained model or a custom model.
- Configure inference parameters such as batch size, number of concurrent users, etc.
- Click "Start Reasoning" and the platform will automatically perform the reasoning task.
- View inference results and performance metrics.
- high performance computing::
- Select the High Performance Computing module.
- Upload the data file to be calculated.
- Configure the parameters of the computation task, such as the number of cores, memory size, etc.
- Click on "Start Calculation" and the platform will automatically carry out the calculation task.
- View calculation results and performance metrics.
- model training::
- Select the Model Training module.
- Upload the training dataset.
- Choose either a pre-trained model or a custom model.
- Configure training parameters such as learning rate, number of training rounds, etc.
- Click "Start Training" and the platform will automatically train the model.
- View training results and model performance.
The Cerebras platform has a simple and intuitive interface that makes it easy for users to get started. The platform also provides detailed documentation and technical support to help users solve problems encountered during use.
