Zephyr 7B Beta 128K is an open-source language model by CallComply. Features: 7b LLM, VRAM: 14.4GB, Context: 128K, License: mit, LLM Explorer Score: 0.13, Arc: 58.3, HellaSwag: 81, MMLU: 53.6, GSM8K: 13.
Zephyr 7B Beta 128K Parameters and Internals
| Model Type | |
| Additional Notes | | Model can produce problematic outputs when prompted as it is not aligned with human preferences for safety. |
|
| Supported Languages | |
| Training Details |
| Data Sources: | | UltraChat dataset, openbmb/UltraFeedback |
|
| Methodology: | | Direct Preference Optimization (DPO) |
|
| Model Architecture: | | 7B parameter GPT-like model |
|
|
| Input Output |
| Accepted Modalities: | |
| Output Format: | |
|
| Release Notes |
| Version: | |
| Notes: | | Zephyr-7B-Ξ² is the second model in the Zephyr series and demonstrates improved performance using Direct Preference Optimization. |
|
|
|
Best Alternatives to Zephyr 7B Beta 128K
Note: green Score (e.g. "73.2") means that the model is better than CallComply/zephyr-7b-beta-128k.