Lamini

🎯 Aiming for 90%+ accuracy on your Text-to-SQL agent, but can't get past 50%? With our proven methodology, our customers have cracked the code and hit 9s of accuracy! We're spilling the tea 🍵 in our upcoming webinar. Bring your toughest Text-to-SQL questions—we’ve got answers! 💪 🎯 Build high-accuracy Text-to-SQL BI agents 📅 March 20, 2025 🕘 10:00 - 10:45 AM PT 🔗 Register here today: bit.ly/41qIycU

Join us for a webinar on building Text-to-SQL BI agents. We’ll show how to finetune any open LLM to reach 90%+ accuracy. Register now bit.ly/41qIycU 🎯 Build high-accuracy Text-to-SQL BI agents 📅 March 20, 2025 🕘 10:00 - 10:45 AM PT

🙌Introducing Memory RAG—a simpler approach to RAG that leverages embed-time compute to create more intelligent, validated data representations. Build mini-agents with a simple prompt. Get the paper: hubs.la/Q0333d5c0

Thanks @DeepLearningAI for featuring us in The Batch! deeplearning.ai/the-batch/a-me…

@nooriefyi Couldn't agree more!

@LaminiAI intent classification is the key to unlocking truly conversational ai

Have you seen our Classifier Agent Toolkit 😺 demo yet? Learn how to use our SDK to build a highly accurate Classifier Agent for a customer service chatbot. The agent categorizes customer interactions by intent so it can respond appropriately. You can run multiple evaluations until you reach your desired level of accuracy. youtube.com/watch?v=-wadT3…

I'm so excited to launch @LaminiAI’s Classifier Agent Toolkit, aka. CAT! 🚀🐱 CAT hunts & tags the important signals 🐭 in a vast amount of data — so devs can easily create agentic classifiers. ❌ Manual data labeling ❌ Large, slow general LLM calls that can only handle 20-30 categories with mid accuracy ✅ CAT has helped our customers tag 2,000 pages across 1,000 categories in just 3.6 seconds with 99.9% accuracy. Dev time? A few hours to a few days. Hallucinations? Approaching zero. Meow. Some common agentic classifiers with CAT: ◽️Customer service agents that extract user intent ◽️Finding high severity tickets, so your teams can prioritize urgent issues ◽️Triage legacy application code based on importance, to prioritize development ◽️Analyze sentiment in earnings calls, reviews, posts, surveys, etc. More on it 👉 lamini.ai/classifier Demo from one of our amazing architects Scott youtube.com/watch?v=-wadT3… Happy holidays, hope you like our gift 🎁 Reach out anytime to fill our inbox with cheer at info@lamini.ai (we read, we respond!) This was a huge effort by the entire Lamini team 🎀

@realSharonZhou 😻😻😻😻😻😻😻😻😻

🎁 Our new Classifier Agent Toolkit (CAT 🐱) is here! No more extensive manual data labeling or heavy ML systems. 😻 Build classifier agents that can quickly categorize large volumes of data at 95%+ accuracy / 400k token throughput in under 2 seconds. Watch the demo and get the link to the docs and repo here: lamini.ai/classifier-age…

🙌 Our new Enterprise Guide to Fine-Tuning is out! If you can't get above 40-50% accuracy with RAG, fine-tuning might be the answer. Learn the basics of fine-tuning and specific applications and use cases. bit.ly/495Q7c9

.@realSharonZhou recently spoke at @Aurecon's #ExemplarForum2024 on high-ROI use cases for LLMs and overcoming key challenges in AI deployment, including poor model quality, hallucinations, costs, and security. Watch the video here: youtube.com/watch?v=gLXT4l…

🎉🎉🎉 Excited to announce our new pay-as-you-go offering, Lamini On-Demand. Get $300 in free credit to run your tuning and inference jobs on our high-performance GPU cluster. Happy tuning! lamini.ai/blog/lamini-on…

@StanTechAddict @GregoryDiamos Great idea @StanTechAddict!

@LaminiAI @GregoryDiamos Certainly one of the best article on LLM inference link to hardware capabilities. May I recommend a part 2 showing performance capabilities with different inference tools (llamacpp, vLLM..) and configuration (CUDA, Vulkan, SYCL, ROCm…)?

LLM inference frameworks have hit the “memory wall”, which is a hardware imposed speed limit on memory bound code. Is it possible to tear down the memory wall? @GregoryDiamos explains how it works in his new technical blog post. lamini.ai/blog/evaluate-…

Go from AI novice to fine-tuning wiz with our Improving Accuracy of LLM Applications course with @DeepLearningAI + @asangani7. Here's one student's experience getting to 96% accuracy on factual data in just 3 iterations. lamini.ai/blog/llm-accur…

Vertical vs. horizontal AI use cases? GitHub Copilot started vertical and crossed over into horizontal applications. Low latency + accuracy were key! Thanks for the great discussion @gajenkandiah and @Hitachi! youtube.com/watch?v=4Wn-rE…

Like many startups, our tech is possible because of access to open source LLMs. @realSharonZhou @matthew_d_white @starlordxie and @pentagoniac recently discussed the importance of an open ecosystem and implications of SB 1047. Thanks to @AIatMeta and @cerebral_valley for hosting and bringing awareness to SB 1047!

🆕 New course on @DeepLearningAI: Improving Accuracy of LLM Applications ➡️ go.fb.me/zfwvd8 Created in collaboration with DL, Meta & @LaminiAI, this free course covers topics like evaluation frameworks, instruction & memory fine-tuning, LoRA + training data generation.

🎉We're excited to share how @LaminiAI, a leading AI cloud service provider, is leveraging our Supermicro GPU Servers to support their large-scale #LLM tuning and generative #AI models. Read our Success Story: hubs.la/Q02LtN5D0 #GenAI #datacenter #GPU #supermicro

Learn a development pattern to systematically improve the accuracy and reliability of LLM applications in our new short course, Improving Accuracy of LLM Applications, built in partnership with @LaminiAI and @Meta, and taught by Lamini’s CEO @realSharonZhou, and Meta’s Senior Director of Partner Engineering, @asangani7. (Disclosure: I am an investor in Lamini.) The path to tuning an LLM application can be complex. In this course, you'll learn a systematic sequence of steps for improving accuracy by reducing hallucinations: - Create an evaluation dataset to measure model accuracy - Add prompt engineering and self-reflection - Fine-tune your model including "memory-tuning" which is a new method of embedding facts in an LLM Using the Llama 3-8B parameter model, you will: - Build a text-to-SQL agent with a custom schema and simulate situations where it hallucinates - Understand the difference between instruction fine-tuning, which gives pre-trained LLMs instructions to follow, and memory fine-tuning - See how Performance-Efficient Fine-tuning (PEFT) techniques like Low-Rank Adaptation (LoRA) reduce training time by 100x and Mixture of Memory Experts (MoME) reduces it even further I appreciate Meta releasing the Llama's family of open models -- this course gives an example of the unique type of work that developers can do with such models. Please sign up here: deeplearning.ai/short-courses/…