File size: 2,459 Bytes
72f212d 43a9a1c 72f212d 43a9a1c edd2996 43a9a1c 7dcdad7 43a9a1c 18a9272 43a9a1c 8578384 43a9a1c 8578384 43a9a1c 8578384 43a9a1c 8578384 43a9a1c 8578384 43a9a1c 8578384 43a9a1c 8578384 43a9a1c cd4953a 43a9a1c |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 |
---
license: other
license_name: yi-license
license_link: https://huggingface.co/01-ai/Yi-34B/blob/main/LICENSE
language:
- en,
pipeline_tag: conversational
---
<p align="center">
<img src="https://cdn-uploads.huggingface.co/production/uploads/644ba0c76ebb3ebf7264dbe9/PWn9I-0XH7kSP_YXcyxIg.png" width="400"/>
</p>
---
# This is a retired model since it was merged with a Capybara, which has been trained wrong with a missing eos_token. Check out the new models: [1](https://huggingface.co/mlinmg/SG-Raccoon-Yi-200k-2.0?text=Hi.) | [2](https://huggingface.co/mlinmg/GS-Raccoon-Yi-200k-2.0?text=Hey+my+name+is+Thomas%21+How+are+you%3F)
# SG Raccoon 55B
The first 55B auto-regressive causal LM created by combining 2x finetuned [Yi 34b](https://huggingface.co/01-ai/Yi-34B) with *200K context* into one.
# Prompting Format
```
SYSTEM: <ANY SYSTEM CONTEXT>
USER:
ASSISTANT:
```
# Merge process
The models used in the merge are [Tess-M-v1.3](https://huggingface.co/migtissera/Tess-M-v1.3/) and [Nous-Capybara-34B](https://huggingface.co/NousResearch/Nous-Capybara-34B).
The layer ranges used are as follows:
```yaml
- model: migtissera/Tess-M-v1.3
layer_range: [0, 14]
- model: NousResearch/Nous-Capybara-34B
layer_range: [7, 21]
- model: migtissera/Tess-M-v1.3
layer_range: [15, 29]
- model: NousResearch/Nous-Capybara-34B
layer_range: [22, 36]
- model: migtissera/Tess-M-v1.3
layer_range: [30, 44]
- model: NousResearch/Nous-Capybara-34B
layer_range: [37, 51]
- model: migtissera/Tess-M-v1.3
layer_range: [45, 59]
```
# Tips
Being a Yi model, try disabling the BOS token and/or running a lower temperature with MinP (and no other samplers) if output doesn't seem right. Yi tends to run "hot" by default.
Sometimes the model "spells out" the stop token as </s> like Capybara, so you may need to add </s> as an additional stopping condition.
# Benchmarks
Coming soon.
# Acknowledgements
- Special thanks to [MSS](https://milanosamplesale.com/) for sponsoring this project
- [@chargoddard](https://huggingface.co/chargoddard) for developing the framework used to merge the model - [mergekit](https://github.com/cg123/mergekit).
- Great thanks to [@Undi95](https://huggingface.co/Undi95) for helping figuring out model merge options
- Also credits to the [01-ai](https://huggingface.co/01-ai) team for their amazing models
- This merged model is inspired by [Goliath 120B](https://huggingface.co/alpindale/goliath-120b) |