Apollo🎖️
Collection
Experimental models
•
3 items
•
Updated
Uploaded model
- Developed by: NotAiLOL
- License: apache-2.0
- Finetuned from model : NotAiLOL/Apollo-7b-Experimental
This mistral model was trained 2x faster with Unsloth and Huggingface's TRL library.
Model detail
Prompt format
The model uses the alpaca format:
Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request.
### Instruction:
{}
### Input:
{}
### Response:
{}
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 71.81 |
| AI2 Reasoning Challenge (25-Shot) | 64.85 |
| HellaSwag (10-Shot) | 85.50 |
| MMLU (5-Shot) | 63.93 |
| TruthfulQA (0-shot) | 63.52 |
| Winogrande (5-shot) | 84.06 |
| GSM8k (5-shot) | 68.99 |
- Downloads last month
- 574
Finetuned from
Dataset used to train NotAiLOL/Apollo-7b-orpo-Experimental
Collection including NotAiLOL/Apollo-7b-orpo-Experimental
Evaluation results
- normalized accuracy on AI2 Reasoning Challenge (25-Shot)test set Open LLM Leaderboard64.850
- normalized accuracy on HellaSwag (10-Shot)validation set Open LLM Leaderboard85.500
- accuracy on MMLU (5-Shot)test set Open LLM Leaderboard63.930
- mc2 on TruthfulQA (0-shot)validation set Open LLM Leaderboard63.520
- accuracy on Winogrande (5-shot)validation set Open LLM Leaderboard84.060
- accuracy on GSM8k (5-shot)test set Open LLM Leaderboard68.990