Tailoring Small Language Models
for Enterprise Use Cases
Julien Simon, Chief Evangelist
julien@arcee.ai
linkedin.com/in/juliensimon
youtube.com/juliensimonfr
Why rent your AI models
when you can own them?
🔒
Increase privacy
and compliance
👔
Tailor models to
your use cases
Maximize ROI
📈
Right-size
cost-performance
🚗
Arcee.ai - The Open SLM leader
State-of-the-art tailoring stack
Spectrum (continuous pre-training), MergeKit (merging), DistilKit (distillation), EvolKit (dataset improvement)
Best-in-class models based on open architectures
Hugging Face OpenLLM Leaderboard benchmarks
Qwen2 72B
🥇
Best Arabic model
Llama 3.1 70B
🥇
Best 70B model
Qwen2 1.5B
🥇
Best 1.5B model
Llama 3.1 8B
🥇
Best 8B model
A modern model adaptation workflow
Pretrained
model
Domain-
adapted
model
Instruction-
tuned model
Aligned
model
Alignment
Merging
instruction-following
behavior
Instruction-
tuned model
Merging
domain
knowledge
Domain-
adapted
model
Merging
alignment
behavior
Aligned
model
Merging steps can be combined, e.g., merge with a domain-adapted and aligned model
📄📄📄
Unlabeled
domain dataset
📄📄📄
Preference dataset
📄📄📄
Q&A dataset
Continuous
pre-training
(CPT)
Instruction
fine-tuning
(IFT)
Spectrum DPO
LoRA
EvolKit
Try it at supernova.arcee.ai
Book a demo at www.arcee.ai/book-a-demo
Deploy it in one click on the AWS Marketplace
Bye bye closed models
AI is changing all businesses.
Make it yours and own it.
Julien Simon, Chief Evangelist
julien@arcee.ai

Tailoring Small Language Models for Enterprise Use Cases

  • 1.
    Tailoring Small LanguageModels for Enterprise Use Cases Julien Simon, Chief Evangelist julien@arcee.ai linkedin.com/in/juliensimon youtube.com/juliensimonfr
  • 2.
    Why rent yourAI models when you can own them? 🔒 Increase privacy and compliance 👔 Tailor models to your use cases Maximize ROI 📈 Right-size cost-performance 🚗
  • 4.
    Arcee.ai - TheOpen SLM leader State-of-the-art tailoring stack Spectrum (continuous pre-training), MergeKit (merging), DistilKit (distillation), EvolKit (dataset improvement) Best-in-class models based on open architectures Hugging Face OpenLLM Leaderboard benchmarks Qwen2 72B 🥇 Best Arabic model Llama 3.1 70B 🥇 Best 70B model Qwen2 1.5B 🥇 Best 1.5B model Llama 3.1 8B 🥇 Best 8B model
  • 5.
    A modern modeladaptation workflow Pretrained model Domain- adapted model Instruction- tuned model Aligned model Alignment Merging instruction-following behavior Instruction- tuned model Merging domain knowledge Domain- adapted model Merging alignment behavior Aligned model Merging steps can be combined, e.g., merge with a domain-adapted and aligned model 📄📄📄 Unlabeled domain dataset 📄📄📄 Preference dataset 📄📄📄 Q&A dataset Continuous pre-training (CPT) Instruction fine-tuning (IFT) Spectrum DPO LoRA EvolKit
  • 6.
    Try it atsupernova.arcee.ai Book a demo at www.arcee.ai/book-a-demo Deploy it in one click on the AWS Marketplace
  • 7.
    Bye bye closedmodels AI is changing all businesses. Make it yours and own it. Julien Simon, Chief Evangelist julien@arcee.ai