---
datasets:
  - ewof/koishi-instruct-metharme
---

## Base Model

native fine tune of togethercomputer/RedPajama-INCITE-Base-3B-v1

## Prompting

The current model version has been trained on prompts using three different roles, which are denoted by the following tokens: `<|system|>`, `<|user|>` and `<|model|>`.

The `<|system|>` prompt can be used to inject out-of-channel information behind the scenes, while the `<|user|>` prompt should be used to indicate user input. The `<|model|>` token should then be used to indicate that the model should generate a response. These tokens can happen multiple times and be chained up to form a conversation history.

## Benchmarks

trained on commit 96baf32 of the linked dataset

![koishi_instruct_3b_v3_benchmarks.png](https://cdn.discordapp.com/attachments/1003543209378578442/1116276126449418300/image.png)

# [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_ewof__koishi-instruct-3b)

| Metric                | Value                     |
|-----------------------|---------------------------|
| Avg.                  | 33.48   |
| ARC (25-shot)         | 40.96          |
| HellaSwag (10-shot)   | 64.54    |
| MMLU (5-shot)         | 26.58         |
| TruthfulQA (0-shot)   | 31.65   |
| Winogrande (5-shot)   | 64.09   |
| GSM8K (5-shot)        | 1.14        |
| DROP (3-shot)         | 5.41         |