OpenInstruct Mistral-7B
1st among commercially-usable 7B models on the Open LLM Leaderboard!*
This is mistralai/Mistral-7B-v0.1 finetuned on VMware/open-instruct.
Quantized to FP16 and released under the Apache-2.0 license by myself.
Compute generously provided by Higgsfield AI.
Prompt format: Alpaca
Below is an instruction that describes a task. Write a response that appropriately completes the request.
### Instruction:
[your instruction goes here]
### Response:
Recommended preset:
- temperature: 0.2
- top_k: 50
- top_p 0.95
- repetition_penalty: 1.1
*as of 21 Nov 2023. "commercially-usable" includes both an open-source base model and a non-synthetic open-source finetune dataset. updated leaderboard results available here.
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 63.64 |
| AI2 Reasoning Challenge (25-Shot) | 59.73 |
| HellaSwag (10-Shot) | 82.77 |
| MMLU (5-Shot) | 60.55 |
| TruthfulQA (0-shot) | 48.76 |
| Winogrande (5-shot) | 79.56 |
| GSM8k (5-shot) | 50.49 |
- Downloads last month
- 3,111
Dataset used to train monology/openinstruct-mistral-7b
Evaluation results
- normalized accuracy on AI2 Reasoning Challenge (25-Shot)test set Open LLM Leaderboard59.730
- normalized accuracy on HellaSwag (10-Shot)validation set Open LLM Leaderboard82.770
- accuracy on MMLU (5-Shot)test set Open LLM Leaderboard60.550
- mc2 on TruthfulQA (0-shot)validation set Open LLM Leaderboard48.760
- accuracy on Winogrande (5-shot)validation set Open LLM Leaderboard79.560
- accuracy on GSM8k (5-shot)test set Open LLM Leaderboard50.490