File size: 1,837 Bytes
0626c8d 8bceaf4 2cbe56e 0626c8d 8bceaf4 2cbe56e |
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 |
---
tags:
- merge
- mergekit
- lazymergekit
- wandb/gemma-2b-zephyr-dpo
- MAISAAI/gemma-2b-coder
base_model:
- wandb/gemma-2b-zephyr-dpo
- MAISAAI/gemma-2b-coder
license: apache-2.0
---
# CodeGemma-2B-Slerp
CodeGemma-2B-Slerp is a merge of the following models using [LazyMergekit](https://colab.research.google.com/drive/1obulZ1ROXHjYLn6PPZJwRR6GzgQogxxb?usp=sharing):
* [wandb/gemma-2b-zephyr-dpo](https://huggingface.co/wandb/gemma-2b-zephyr-dpo)
* [MAISAAI/gemma-2b-coder](https://huggingface.co/MAISAAI/gemma-2b-coder)
Special thanks to Charles Goddard for the quick implementation!
## 🏆 Evaluation
### Coming Soon
## 🧩 Configuration
```yaml
slices:
- sources:
- model: wandb/gemma-2b-zephyr-dpo
layer_range: [0, 18]
- model: MAISAAI/gemma-2b-coder
layer_range: [0, 18]
merge_method: slerp
base_model: wandb/gemma-2b-zephyr-dpo
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
dtype: bfloat16
```
## 💻 Usage
```python
!pip install -qU transformers accelerate
from transformers import AutoTokenizer
import transformers
import torch
model = "johnsnowlabs/CodeGemma-2B-Slerp"
messages = [{"role": "user", "content": "What is a large language model?"}]
tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
"text-generation",
model=model,
torch_dtype=torch.float16,
device_map="auto",
)
outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])
``` |