Update README.md
Browse files
README.md
CHANGED
|
@@ -1,3 +1,33 @@
|
|
| 1 |
---
|
| 2 |
license: cc-by-nc-4.0
|
|
|
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: cc-by-nc-4.0
|
| 3 |
+
pipeline_tag: text-generation
|
| 4 |
---
|
| 5 |
+
|
| 6 |
+
# Pirouette-7b
|
| 7 |
+
|
| 8 |
+
This is a gradient SLERP merge of Eric Hartford's [dolphin-2.1-mistral-7b](https://huggingface.co/ehartford/dolphin-2.1-mistral-7b) merged with Undi and IkariDev's [Noromaid-v0.1.1-13b](https://huggingface.co/NeverSleep/Noromaid-13b-v0.1.1). The merge is weighted towards Dolphin.
|
| 9 |
+
|
| 10 |
+
The prompt format is Alpaca. You can use the standard format as shown, but for best results, I strongly recommend customizing the system prompt to your specific needs.
|
| 11 |
+
|
| 12 |
+
```
|
| 13 |
+
You are Dolphin, a helpful AI assistant.
|
| 14 |
+
|
| 15 |
+
### Instruction:
|
| 16 |
+
{YOUR MESSAGE HERE}
|
| 17 |
+
|
| 18 |
+
### Response:
|
| 19 |
+
{BOT MESSAGE HERE}
|
| 20 |
+
|
| 21 |
+
|
| 22 |
+
```
|
| 23 |
+
|
| 24 |
+
### Misc. information
|
| 25 |
+
- BOS token is `<s>`
|
| 26 |
+
- EOS token is `</s>`
|
| 27 |
+
- Native context length is `8192`
|
| 28 |
+
- Functional context length extended to 32768 via RoPE with decreased perplexity, see [here](https://github.com/ggerganov/llama.cpp/issues/3867#issuecomment-1789054205)
|
| 29 |
+
- Base model is Mistral v0.1
|
| 30 |
+
|
| 31 |
+
### Thanks
|
| 32 |
+
- Thanks to [Eric Hartford](https://ko-fi.com/erichartford) for Dolphin
|
| 33 |
+
- Thanks to [Undi](https://ko-fi.com/undiai) and [IkariDev](https://ikaridevgit.github.io/) for Noromaid
|