DeciLM-7B

DeciLM-7B is a 7.04 billion parameter decoder-only text generation model, released under the Apache 2.0 license. At the time of release, DeciLM-7B is the top-performing 7B base language model on the Open LLM Leaderboard. With support for an 8K-token sequence length, this highly efficient model uses variable Grouped-Query Attention (GQA) to achieve a superior balance between accuracy and computational efficiency. The model's architecture was generated using Deci's proprietary Neural Architecture Search technology, AutoNAC.

Model Details

Model Description

Deci developed and released the DeciLM-7B language model, a pre-trained, high-efficiency text generation model with 7 billion parameters. DeciLM-7B is not only the most accurate 7B base model, but it also outpaces all models in its class with a throughput that is up to 4.4x that of Mistral-7B's. An instruct version DeciLM-7B-instruct has also been released.

• Developed by: Deci
• Model type: DeciLM is an auto-regressive language model using an optimized transformer decoder architecture that includes variable Grouped-Query Attention.
• Language(s) (NLP): English
• License: Apache 2.0

Model Architecture

Parameters	Layers	Heads	Sequence Length	GQA num_key_value_heads*
7.04 billion	32	32	8192	Variable

*AutoNAC was employed to optimize the selection of the GQA num_key_value_heads for each layer.

Model Sources

• Blog: DeciLM-7B Technical Blog
• Demo: DeciLM-7B-instruct Demo
• Finetuning Notebook: DeciLM-7B Finetuning Notebook
• Text Generation Notebook: DeciLM-7B-instruct Text Generation Notebook

Uses

The model is intended for commercial and research use in English and can be fine-tuned for various tasks and languages.

How to Get Started with the Model

Use the code below to get started with the model.

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "Deci/DeciLM-7B"
device = "cuda" # for GPU usage or "cpu" for CPU usage

tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype="auto", trust_remote_code=True).to(device)

inputs = tokenizer.encode("In a shocking finding, scientists discovered a herd of unicorns living in", 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]))

# The model can also be used via the text-generation pipeline interface
from transformers import pipeline
generator = pipeline("text-generation", "Deci/DeciLM-7B", torch_dtype="auto", trust_remote_code=True, device=device)
outputs = generator("In a shocking finding, scientists discovered a herd of unicorns living in", max_new_tokens=100, do_sample=True, top_p=0.95)
print(outputs[0]["generated_text"])

Evaluation

Below are DeciLM-7B and DeciLM-7B-instruct's Open LLM Leaderboard results.

Model	Average	ARC	HellaSwag	MMLU	TruthfulQA	Winogrande	GSM8K
DecilLM-7B	61.55	59.39	82.51	59.76	40.33	79.95	47.38
DecilLM-7B-instruct	63.19	61.01	82.37	60.24	49.75	79.72	46.02

Runtime Benchmarks

Inference Tool	Hardware	Prompt length	Generation length	Generated tokens/sec	Batch Size	Number of Prompts
HuggingFace (PyTorch)	A100 (SXM4-80GB-400W)	512	512	1174	352	352
HuggingFace (PyTorch)	A100 (SXM4-80GB-400W)	2048	2048	328	72	72
Infery-LLM	A100 (SXM4-80GB-400W)	512	512	4559	1024	4096
Infery-LLM	A100 (SXM4-80GB-400W)	2048	2048	3997	512	2048
Infery-LLM	A10	512	512	1345	128	512
Infery-LLM	A10	2048	2048	599	32	128

• In order to replicate the results of the Hugging Face benchmarks, you can use this code example.
• Infery-LLM, Deci's inference engine, features a suite of optimization algorithms, including selective quantization, optimized beam search, continuous batching, and custom CUDA kernels. To explore the capabilities of Infery-LLM, schedule a live demo.

Ethical Considerations and Limitations

DeciLM-7B is a new technology that comes with inherent risks associated with its use. The testing conducted so far has been primarily in English and does not encompass all possible scenarios. Like those of all large language models, DeciLM-7B's outputs are unpredictable, and the model may generate responses that are inaccurate, biased, or otherwise objectionable. Consequently, developers planning to use DeciLM-7B should undertake thorough safety testing and tuning designed explicitly for their intended applications of the model before deployment.

How to Cite

Please cite this model using this format.

@misc{DeciFoundationModels,
title = {DeciLM-7B},
author = {DeciAI Research Team},
year = {2023}
url={https://huggingface.co/Deci/DeciLM-7B},
}