Compare Base Models
Types of Base Models #
There are typically three types of base models: generalized, chat, and instruct models.
- Generalized Models: These models have been fine-tuned to be able to perform a wide range of tasks.
- Chat Models: These models have been fine-tuned to be able to interact with users via chat.
- Instruct Models: These models have been fine-tuned to be able to follow prompted instructions.
Available Base Models #
Generalized Models #
Use generalized models when you intend to provide an inferencing service for a wide range of tasks. They are also great for testing the capabilities of that model’s family.
| Model Name | Parameters |
|---|---|
| Llama-2-7b | 7 Billion |
| Llama-2-13b | 13 Billion |
| Llama-2-70b | 70 Billion |
| falcon-7b | 7 Billion |
| falcon-40b | 40 Billion |
| mpt-7b | 7 Billion |
| mpt-30b | 30 Billion |
Chat Models #
Use chat models when you intend to provide an inferencing service for conversational AI applications.
| Model Name | Parameters |
|---|---|
| Llama-2-7b-chat | 7 Billion |
| Llama-2-13b-chat | 13 Billion |
| Llama-2-70b-chat | 70 Billion |
| falcon-7b-chat | 7 Billion |
Instruct Models #
Use instruct models when you want to programmatically instruct the model to perform a task, such as data transformation, classification, summarization, and more.
| Model Name | Parameters |
|---|---|
| falcon-40b-instruct | 40 Billion |
| mpt-7b-instruct | 7 Billion |
| mpt-30b-instruct | 30 Billion |