Orca 2 Models by Microsoft Research
About
Orca 2 is a family of small language models (SLMs) created by Microsoft Research, specifically engineered to enhance reasoning capabilities in smaller frameworks. Unlike large language models (LLMs) that heavily focus on size, Orca 2 demonstrates that smaller models can achieve performance comparable to or even exceeding larger models through innovative techniques. These models utilize instruction tuning, explanation tuning, and a unique method that omits conventional system prompts for more strategic reasoning. They are available in 7 billion and 13 billion parameter versions, both fine-tuned from LLaMA 2 base models with high-quality synthetic data. Orca 2 excels in reading comprehension, math problem-solving, and text summarization while being openly accessible for research. Despite its robust abilities, it still encompasses limitations such as potential biases and risks of generating inaccurate content 148.
Current Variants
Use-when guidance is derived from seed capabilities, context, release, and replacement fields.
Use when the workload needs 4k context and 13B parameters.
Use when the workload needs 4k context and 7B parameters.
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| Orca 2 13B | Use when the workload needs 4k context and 13B parameters. | 2023-11 | 4k context13B parameters | Current |
| Orca 2 7B | Use when the workload needs 4k context and 7B parameters. | 2023-11 | 4k context7B parameters | Current |
Release Timeline
1 release groupSpecifications(2 models)
| Model | Released | Context | Parameters |
|---|---|---|---|
| Orca 2 13B | 2023-11 | 4k | 13B |
| Orca 2 7B | 2023-11 | 4k | 7B |
Available From(1 provider)
Pricing
| Model | Provider | Input / 1M | Output / 1M | Type |
|---|---|---|---|---|
| Orca 2 7B | Microsoft Foundry | $0.52 | $0.67 | Provisioned |
| Orca 2 13B | Microsoft Foundry | $0.81 | $0.94 | Provisioned |
Frequently Asked Questions
- What is Orca 2 used for?
- Orca 2 is used for math-heavy prompts. The family description and listed model capabilities point to those workloads as the best fit.
- How does Orca 2 compare to Harrier?
- Orca 2 by Microsoft Research is strongest where you need math-heavy prompts, while Harrier by Microsoft Research is the closest related family to check for embedding. Orca 2 has 2 listed variants and reaches up to 4k context, while Harrier reaches up to 33k context, so compare the specs and pricing tables before choosing a production model.
- Which Orca 2 model should I use?
- For the lowest listed input price, start with Orca 2 7B through Microsoft Foundry at $0.52/1M input tokens. For the most capable/latest local choice, evaluate Orca 2 13B with 4k context.
