🚀 DeciLM 6B-Instruct
DeciLM 6B-Instruct is a model designed for short-form instruction following. It addresses the need for efficient instruction processing in English language tasks, offering both commercial and research value.
🚀 Quick Start
Use the following code to start using the model:
# pip install -q transformers
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
checkpoint = "Deci/DeciLM-6b-instruct"
device = "cuda" # for GPU usage or "cpu" for CPU usage
tokenizer = AutoTokenizer.from_pretrained(checkpoint)
model = AutoModelForCausalLM.from_pretrained(checkpoint, torch_dtype=torch.bfloat16, trust_remote_code=True).to(device)
inputs = tokenizer.encode("How do I make french toast? Think through it step by step", return_tensors="pt").to(device)
outputs = model.generate(inputs, max_new_tokens=100, do_sample=True, top_p=0.95)
print(tokenizer.decode(outputs[0]))
✨ Features
- Commercial and Research Use: Suitable for both commercial applications and research purposes in English, and can be fine - tuned for other languages.
- Fast Training: Trained using advanced proprietary methodologies with the SlimPijamas dataset, enabling fast training.
- Fine - Tuned: Further fine - tuned on a subset of the OpenOrca dataset to create DeciLM-6B-Instruct.
📦 Installation
The installation command is included in the quick start code:
# pip install -q transformers
📚 Documentation
Model Information
Property |
Details |
Model Type |
DeciLM is an auto - regressive language model using an optimized transformer decoder architecture that includes variable Grouped - Query Attention. |
Training Data |
SlimPajamas dataset for initial training, and a subset of the OpenOrca dataset for fine - tuning. |
License |
Llama 2 Community License Agreement with an extention of Deci regarding hosting service providers. |
Model Sources
Uses
The model is intended for commercial and research use in English and can be fine - tuned for use in other languages.
Training Details
DeciLM 6B was trained using the SlimPijamas dataset with advanced proprietary methods for fast training. It was further fine - tuned on a subset of the OpenOrca dataset to create DeciLM-6B-Instruct.
Evaluation
Average |
ARC Challenge* |
ARC Easy* |
BoolQ |
HellaSwag* |
LAMBDA OpenAI |
OpenBookQA |
PIQA |
TruthfulQA |
Winogrande |
62.01 |
44.43 |
70.58 |
77.34 |
74.57 |
70.1 |
33 |
77.52 |
43.89 |
67.64 |
Accuracy - norm score* |
|
|
|
|
|
|
|
|
|
Runtime Benchmarks
Inference Tool/Hardware |
A10 (tokens/sec) |
PyTorch |
652.49 |
Infery LLM |
2,029.6 |
- Throughput (tokens/sec) - Measured with optimal batch - PyTorch BS 64, Infery LLM BS 128
- In order to replicate the results of the PyTorch benchmark, use this code example
Disclaimer
DeciLM 6B - Instruct has not been aligned for safety or trained using RLHF.
How to Cite
Please cite this model using the following format:
@misc{DeciFoundationModels,
title = {DeciLM 6B Instruct},
author = {DeciAI Research Team},
year = {2023}
url={[https://huggingface.co/Deci/DeciLM-6b-instruct](https://huggingface.co/Deci/DeciLM-6b-instruct)},
}