File size: 2,705 Bytes
4587814 d92f86e 4587814 d92f86e ff15ee2 d92f86e ff15ee2 da36c21 d92f86e ff15ee2 d92f86e da36c21 d92f86e ff15ee2 d92f86e ff15ee2 d92f86e |
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 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 |
---
language:
- en
- ko
pipeline_tag: text-generation
license: cc-by-nc-sa-4.0
---
# **SOLAR-tail-10.7B-Merge-v1.0**
## Model Details
**Model Developers** Kyujin Han (kyujinpy)
**Method**
Using [Mergekit](https://github.com/cg123/mergekit).
- [upstage/SOLAR-10.7B-v1.0](https://huggingface.co/upstage/SOLAR-10.7B-v1.0)
- [Yhyu13/LMCocktail-10.7B-v1](Yhyu13/LMCocktail-10.7B-v1)
**Merge config**
```
slices:
- sources:
- model: upstage/SOLAR-10.7B-v1.0
layer_range: [0, 48]
- model: Yhyu13/LMCocktail-10.7B-v1
layer_range: [0, 48]
merge_method: slerp
base_model: upstage/SOLAR-10.7B-v1.0
parameters:
t:
- filter: self_attn
value: [0, 0.5, 0.3, 0.7, 1]
- filter: mlp
value: [1, 0.5, 0.7, 0.3, 0]
- value: 0.5 # fallback for rest of tensors
tokenizer_source: union
dtype: float16
```
# **Model Benchmark**
## Open Ko leaderboard
- Follow up as [Ko-link](https://huggingface.co/spaces/upstage/open-ko-llm-leaderboard).
| Model | Average | ARC | HellaSwag | MMLU | TruthfulQA | Ko-CommonGenV2 |
| --- | --- | --- | --- | --- | --- | --- |
| PracticeLLM/SOLAR-tail-10.7B-Merge-v1.0 | NaN | NaN | NaN | NaN | NaN | NaN |
| jjourney1125/M-SOLAR-10.7B-v1.0 | 55.15 | 49.57 | 60.12 | 54.60 | 49.23 | 62.22 |
| beomi/Yi-Ko-6B | 48.79 | 41.04 | 53.39 | 46.28 | 41.64 | 61.63 |
| mistralai/Mistral-7B-v0.1 | 46.89 | 38.14 | 48.19 | 45.20 | 46.13 | 56.79 |
- Follow up as [En-link](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard).
## lm-evaluation-harness
```
gpt2 (pretrained=PracticeLLM/SOLAR-tail-10.7B-Merge-v1.0), limit: None, provide_description: False, num_fewshot: 0, batch_size: None
| Task |Version| Metric |Value | |Stderr|
|----------------|------:|--------|-----:|---|-----:|
|kobest_boolq | 0|acc |0.5021|± |0.0133|
| | |macro_f1|0.3343|± |0.0059|
|kobest_copa | 0|acc |0.6220|± |0.0153|
| | |macro_f1|0.6217|± |0.0154|
|kobest_hellaswag| 0|acc |0.4380|± |0.0222|
| | |acc_norm|0.5380|± |0.0223|
| | |macro_f1|0.4366|± |0.0222|
|kobest_sentineg | 0|acc |0.4962|± |0.0251|
| | |macro_f1|0.3316|± |0.0113|
```
# Implementation Code
```python
### KO-Platypus
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
repo = "PracticeLLM/SOLAR-tail-10.7B-Merge-v1.0"
OpenOrca = AutoModelForCausalLM.from_pretrained(
repo,
return_dict=True,
torch_dtype=torch.float16,
device_map='auto'
)
OpenOrca_tokenizer = AutoTokenizer.from_pretrained(repo)
```
--- |