WizardCoder Models by WizardLM Team
About
WizardCoder is a family of large language models (LLMs) specialized for code generation and understanding. Developed by fine-tuning models like Llama 2 using the Evol-Instruct method, WizardCoder models have achieved state-of-the-art performance on coding benchmarks like HumanEval. The family includes versions like WizardCoder-15B and WizardCoder-Python-34B, which have outperformed many open-source and even some closed-source coding models. WizardCoder's success is attributed to its unique training dataset that balances simple, complex, and highly complex coding instructions, enabling it to handle a wide range of programming tasks effectively.
Current Variants
Use-when guidance is derived from seed capabilities, context, release, and replacement fields.
Use when the workload needs 16k context and 33B parameters.
Use when the workload needs 100k context, 34B parameters, and structured outputs.
Use when the workload needs 8k context and 15B parameters.
Use when the workload needs 100k context and 13B parameters.
Use when the workload needs 100k context and 7B parameters.
Use when the workload needs 8k context and 3B parameters.
Use when the workload needs 8k context and 1B parameters.
| Model | Use when | Released | Signals | Status |
|---|---|---|---|---|
| WizardCoder 33B | Use when the workload needs 16k context and 33B parameters. | 2024-01 | 16k context33B parameters | Current |
| WizardCoder Python 34B | Use when the workload needs 100k context, 34B parameters, and structured outputs. | 2024-01 | 100k context34B parametersstructured outputs | Current |
| Wizard Coder 15B | Use when the workload needs 8k context and 15B parameters. | 2024-01 | 8k context15B parameters | Current |
| WizardCoder Python 13B | Use when the workload needs 100k context and 13B parameters. | 2024-01 | 100k context13B parameters | Current |
| WizardCoder Python 7B | Use when the workload needs 100k context and 7B parameters. | 2024-01 | 100k context7B parameters | Current |
| WizardCoder 3B | Use when the workload needs 8k context and 3B parameters. | 2024-01 | 8k context3B parameters | Current |
| WizardCoder 1B | Use when the workload needs 8k context and 1B parameters. | 2024-01 | 8k context1B parameters | Current |
Release Timeline
1 release groupSpecifications(7 models)
| Model | Released | Context | Parameters | Structured Outputs |
|---|---|---|---|---|
| WizardCoder 33B | 2024-01 | 16k | 33B | No |
| WizardCoder Python 34B | 2024-01 | 100k | 34B | Yes |
| Wizard Coder 15B | 2024-01 | 8k | 15B | No |
| WizardCoder Python 13B | 2024-01 | 100k | 13B | No |
| WizardCoder Python 7B | 2024-01 | 100k | 7B | No |
| WizardCoder 3B | 2024-01 | 8k | 3B | No |
| WizardCoder 1B | 2024-01 | 8k | 1B | No |
Available From(2 providers)
Pricing
| Model | Provider | Input / 1M | Output / 1M | Type |
|---|---|---|---|---|
| WizardCoder 33B | Replicate API | $0.2 | $1 | Serverless |
| WizardCoder Python 34B | Replicate API | $0.2 | $1 | Serverless |
| WizardCoder Python 34B | Together AI | $0.8 | $0.8 | Serverless |
Frequently Asked Questions
- What is WizardCoder used for?
- WizardCoder is used for coding, structured outputs, and math-heavy prompts. The family description and listed model capabilities point to those workloads as the best fit.
- How does WizardCoder compare to StarCoder?
- WizardCoder by WizardLM Team is strongest where you need coding, while StarCoder by ServiceNow Research is the closest related family to check for coding. WizardCoder has 7 listed variants and reaches up to 100k context, while StarCoder reaches up to 8k context, so compare the specs and pricing tables before choosing a production model.
- Which WizardCoder model should I use?
- For the lowest listed input price, start with WizardCoder 33B through Replicate API at $0.2/1M input tokens. For the most capable/latest local choice, evaluate WizardCoder Python 34B with 100k context and structured outputs.
