File size: 3,149 Bytes
0db91b0 89b7232 7885d2e 89b7232 |
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 72 73 |
---
license: cc-by-nc-4.0
tags:
- merge
---
### Chatty-2x8B
## Description
After some testing, finetuning and multiple merges of Llama-3 LLM models, here is something a little different.
This model is a MoE of 2x Llama-3 model trained on different RP format.
This repo contains FP16 files of Chatty-2x8B.
## The idea
I started with two separate Llama-3-Instruct-8B models, each fine-tuned for specific RP formats.
Here is two simple exemple of how it was trained.
- **Expert 1**: This model is trained to handle RP that requires actions and descriptions between asterisks. For example:
```
*nods* Yes, I understand.
```
- **Expert 2**: This model is fine-tuned for plain text RP where characters’ dialogues and actions are described straightforwardly. For example:
```
Nods. "Yes, I understand."
```
My initial idea was to make a 11B or bigger Llama-3 model, or just make a 2x8B from existing model, but I got some issues, they were not stable enough, even after DPO and FFT on top my frankenmerge/moe of Llama-3, it was not working well enough to release them.
So I just tried the idea of having 2 different RP format trained on 2 separated Llama-3-Instruct-8B, and it worked pretty well!
## The dataset
Based on Lumimaid 8B OAS success I still used the same "balance" between RP and non RP in the dataset, the maximum was 50% non RP data on each side.
RP data was different with some exception, the non RP data was exactly the same, despite that, I can't produce repetition so the double usage of non RP datasets didn't hurt the model in the end.
## Prompt template: Llama3
```
<|begin_of_text|><|start_header_id|>system<|end_header_id|>
{system_prompt}<|eot_id|><|start_header_id|>user<|end_header_id|>
{input}<|eot_id|><|start_header_id|>assistant<|end_header_id|>
{output}<|eot_id|>
```
## Others
Undi: If you want to support us, you can [here](https://ko-fi.com/undiai).
IkariDev: Visit my [retro/neocities style website](https://ikaridevgit.github.io/) please kek
| Tasks |Version| Filter |n-shot| Metric |Value | |Stderr|
|--------------|------:|----------------|-----:|-----------|-----:|---|-----:|
|arc_challenge | 1|none | 0|acc |0.5469|± |0.0145|
| | |none | 0|acc_norm |0.5853|± |0.0144|
|arc_easy | 1|none | 0|acc |0.8308|± |0.0077|
| | |none | 0|acc_norm |0.8258|± |0.0078|
|gsm8k | 3|strict-match | 5|exact_match|0.7149|± |0.0124|
| | |flexible-extract| 5|exact_match|0.7096|± |0.0125|
|hellaswag | 1|none | 0|acc |0.5945|± |0.0049|
| | |none | 0|acc_norm |0.7806|± |0.0041|
|piqa | 1|none | 0|acc |0.7943|± |0.0094|
| | |none | 0|acc_norm |0.7998|± |0.0093|
|truthfulqa_mc2| 2|none | 0|acc |0.5097|± |0.0150|
|winogrande | 1|none | 0|acc |0.7356|± |0.0124| |