|
--- |
|
base_model: |
|
- princeton-nlp/Llama-3-8B-ProLong-64k-Instruct |
|
tags: |
|
- merge |
|
- mergekit |
|
- lazymergekit |
|
- princeton-nlp/Llama-3-8B-ProLong-64k-Instruct |
|
library_name: transformers |
|
license: llama3 |
|
--- |
|
|
|
# Llama-3-15B-64k-Instruct |
|
|
|
I decided to repeat [this](https://huggingface.co/elinas/Llama-3-15B-Instruct-zeroed) merge, but using [64K version of Llama 3 8B](https://huggingface.co/princeton-nlp/Llama-3-8B-ProLong-64k-Instruct). |
|
|
|
This should work with a context up to 64k, but I strongly recommend making a finetune first. |
|
|
|
## 💻 Usage |
|
|
|
```python |
|
!pip install -qU transformers accelerate |
|
|
|
from transformers import AutoTokenizer |
|
import transformers |
|
import torch |
|
|
|
model = "Ttimofeyka/Llama-3-15B-64k-Instruct" |
|
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"]) |
|
``` |
