|
--- |
|
license: cc-by-nc-4.0 |
|
base_model: |
|
- SvalTek/L3-ColdBrew-Astrid |
|
- FPHam/L3-8B-Everything-COT |
|
tags: |
|
- moe |
|
- frankenmoe |
|
- merge |
|
- mergekit |
|
- lazymergekit |
|
- SvalTek/L3-ColdBrew-Astrid |
|
- FPHam/L3-8B-Everything-COT |
|
--- |
|
|
|
# ColdBrew-Oxford |
|
|
|
ColdBrew-Oxford is a Mixture of Experts (MoE) made with the following models using [LazyMergekit](https://colab.research.google.com/drive/1obulZ1ROXHjYLn6PPZJwRR6GzgQogxxb?usp=sharing): |
|
* [SvalTek/L3-ColdBrew-Astrid](https://huggingface.co./SvalTek/L3-ColdBrew-Astrid) |
|
* [FPHam/L3-8B-Everything-COT](https://huggingface.co./FPHam/L3-8B-Everything-COT) |
|
|
|
## 🧩 Configuration |
|
|
|
```yaml |
|
base_model: bunnycore/LLama-3.1-Hyper-Stock |
|
experts_per_token: 2 |
|
gate_mode: hidden |
|
dtype: bfloat16 |
|
experts: |
|
- source_model: SvalTek/L3-ColdBrew-Astrid |
|
positive_prompts: |
|
- "ColdBrew --" |
|
- "Write a story about a lonely robot exploring an abandoned space station." |
|
- "Describe the feeling of standing at the edge of a massive waterfall." |
|
- "Roleplay as a tavern keeper sharing tales with an adventurer." |
|
- "Imagine a bustling marketplace in a desert city and narrate the sights and sounds." |
|
- "[Genres: Science Fiction] |
|
[Tags: humor, old school, sci fi]" |
|
- "[Mode: Interactive Storyteller]" |
|
- "[Mode: DM]" |
|
negative_prompts: |
|
- "Think step by step" |
|
|
|
- source_model: FPHam/L3-8B-Everything-COT |
|
positive_prompts: |
|
- "[OOC:" |
|
- "What are <thinking> and <reflection> blocks used for?" |
|
- "Reflect on why humans lie to each other." |
|
- "What are the most important aspects of good technical documentation?" |
|
- "Analyzeand explain the logic behind a recursive algorithm for sorting data." |
|
- "Think step by step with a logical reasoning and intellectual sense before you provide any response." |
|
negative_prompts: |
|
- "Alex is on a spaceship, standing before a mysterious control panel." |
|
``` |
|
|
|
## 💻 Usage |
|
|
|
```python |
|
!pip install -qU transformers bitsandbytes accelerate |
|
|
|
from transformers import AutoTokenizer |
|
import transformers |
|
import torch |
|
|
|
model = "SvalTek/ColdBrew-Oxford" |
|
|
|
tokenizer = AutoTokenizer.from_pretrained(model) |
|
pipeline = transformers.pipeline( |
|
"text-generation", |
|
model=model, |
|
model_kwargs={"torch_dtype": torch.float16, "load_in_4bit": True}, |
|
) |
|
|
|
messages = [{"role": "user", "content": "Explain what a Mixture of Experts is in less than 100 words."}] |
|
prompt = pipeline.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True) |
|
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"]) |
|
``` |
