File size: 1,850 Bytes
c5d6f36 5fe367d 7f00dda c5d6f36 5fe367d c5d6f36 5fe367d 7f00dda 5fe367d |
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 |
---
tags:
- merge
- mergekit
- lazymergekit
- aaditya/Llama3-OpenBioLLM-8B
base_model:
- aaditya/Llama3-OpenBioLLM-8B
license: llama3
language:
- en
---
# Llama-3-Galen-8B-32k-v1
<img src="https://cdn-uploads.huggingface.co/production/uploads/60c8619d95d852a24572b025/R73wGdZE3GWeF9QZPvruG.jpeg" width="600" />
Llama-3-Galen-8B-32k-v1 is a RoPE scaled, DARE TIES merge of the following models using [LazyMergekit](https://colab.research.google.com/drive/1obulZ1ROXHjYLn6PPZJwRR6GzgQogxxb?usp=sharing):
* [aaditya/Llama3-OpenBioLLM-8B](https://huggingface.co/aaditya/Llama3-OpenBioLLM-8B)
* [johnsnowlabs/JSL-MedLlama-3-8B-v2.0](https://huggingface.co/johnsnowlabs/JSL-MedLlama-3-8B-v2.0)
> **This model is capable of handling a context size of 32K right out of the box, enabled with Dynamic RoPE scaling.**
## 🧩 Configuration
```yaml
models:
- model: johnsnowlabs/JSL-MedLlama-3-8B-v2.0
# No parameters necessary for base model
- model: aaditya/Llama3-OpenBioLLM-8B
parameters:
density: 0.53
weight: 0.5
merge_method: dare_ties
base_model: johnsnowlabs/JSL-MedLlama-3-8B-v2.0
parameters:
int8_mask: true
dtype: bfloat16
```
## 💻 Usage
```python
!pip install -qU transformers accelerate
from transformers import AutoTokenizer
import transformers
import torch
model = "abhinand/Llama-3-Galen-8B-32k-v1"
messages = [{"role": "user", "content": "How long does it take to recover from COVID-19?"}]
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"])
``` |