CrystalMistral-24B
CrystalMistral-24B is a Mixure of Experts (MoE) made with the following models using LazyMergekit:
- eren23/dpo-binarized-NeuralTrix-7B
- macadeliccc/WestLake-7B-v2-laser-truthy-dpo
- Weyaxi/OpenHermes-2.5-neural-chat-v3-2-Slerp
- cognitivecomputations/WestLake-7B-v2-laser
🧩 Configuration
base_model: eren23/dpo-binarized-NeuralTrix-7B
gate_mode: hidden
dtype: bfloat16
experts:
- source_model: eren23/dpo-binarized-NeuralTrix-7B
positive_prompts:
- "Generate a response to a given situation"
- "Explain the concept of climate change"
- source_model: macadeliccc/WestLake-7B-v2-laser-truthy-dpo
positive_prompts:
- "What is the capital of France?"
- "Who wrote the novel 'Pride and Prejudice'?"
- source_model: Weyaxi/OpenHermes-2.5-neural-chat-v3-2-Slerp
positive_prompts:
- "Write a short poem about spring"
- "Design a logo for a tech startup called 'GreenLeaf'"
- source_model: cognitivecomputations/WestLake-7B-v2-laser
positive_prompts:
- "Solve the equation x^2 + 3x - 10 = 0"
- "Calculate the area of a circle with radius 5 units"
💻 Usage
!pip install -qU transformers bitsandbytes accelerate
from transformers import AutoTokenizer
import transformers
import torch
model = "Crystalcareai/CrystalMistral-24B"
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"])