README.md

---
license: apache-2.0
tags:
  - autotrain
  - text-generation
base_model: Locutusque/TinyMistral-248M
datasets:
  - KnutJaegersberg/WizardLM_evol_instruct_V2_196k_instruct_format
widget:
  - text: |-
      ### Instruction
      Find me a list of some nice places to visit around the world.

      ### Response:
  - text: |-
      ### Instruction
      Tell me a story.
      Once upon a time...

      ### Response:
inference:
  parameters:
    max_new_tokens: 32
    repetition_penalty: 1.15
    do_sample: true
    temperature: 0.5
---

# Locutusque's TinyMistral-248M trained on the Evol Instruct dataset using AutoTrain

- Base model: [Locutusque/TinyMistral-248M](https://huggingface.co/Locutusque/TinyMistral-248M)
- Dataset: [KnutJaegersberg/WizardLM_evol_instruct_V2_196k_instruct_format](https://huggingface.co/datasets/KnutJaegersberg/WizardLM_evol_instruct_V2_196k_instruct_format)
- Training: 9.5h under [these parameters](https://huggingface.co/Felladrin/TinyMistral-248M-Evol-Instruct/blob/85723f768c439307c996328d68fe7df2947237ae/training_params.json)

## Recommended Prompt Format

```
<instruction>

### Response:
```

## Recommended Inference Parameters

```yml
repetition_penalty: 1.15
do_sample: true
temperature: 0.5
```