We’re seeing more and more businesses switch from closed to open-source language models. You can fine-tune models like Llama 2 and Mistral to perform better than GPT-4 on particular tasks with a much smaller, cheaper model.
GENERATIVE AI
-
Language Models Drive Growth for Replicate After Llama 2 Release
By
–
Although we got our start in generative images, language models are where we see most of our growth. The week Llama 2 came out was our biggest week of growth ever.
-
Large Businesses Deploy AI Models on Replicate Platform
By
–
Large businesses are building on us too. Unsplash are labelling all the images in their catalog using BLIP. BuzzFeed are… turning your pets into plushies. Character AI, Labelbox, and thousands of other businesses are deploying models on Replicate.
-
Indie hackers turn side projects into million-dollar businesses
By
–
But soon, side projects started to turn into real products. Indie hackers like @levelsio and @dannypostma made apps that generate profile pictures, redecorate your house and create professional headshots. They’re now real businesses making over $1 million annual revenue.
-
Stable Diffusion: From Toy to Open-Source Image Generation Tool
By
–
Last year, Stable Diffusion was released. It was an open-source image generation model that caught the imagination of tinkerers. At the start, it felt like a toy. It was just people tinkering around and seeing what was possible.
-
LangChain Extraction Benchmarking Results and Feedback
By
–
For all these results and more, check out the post and let us know your feedback! https://
blog.langchain.dev/extraction-ben
chmarking/
… -
Grammar-Based Decoding and Model Quality for Structured Output
By
–
Grammar-based decoding (like logit biasing) helps, but isn't a panacea – model quality is also important for generating correct values, even after you have the structure locked down.
-
LangChain Chat Extraction Benchmarks Documentation Guide
By
–
You can try it yourself on your own setup by following the doc! https://
langchain-ai.github.io/langchain-benc
hmarks/notebooks/extraction/chat_extraction.html
… -
Testing LLM Performance with Prompting and Decoding Techniques
By
–
We tested popular models like GPT-4, Claude, Llama, and Yi using different prompting and decoding techniques to see how much they impact performance for things like: – Question classification
– Sentiment analysis
– Nested JSON schema correctness -
GPT-4 and Llama Performance Comparison on Tool-Calling Benchmark
By
–
Unsurprisingly, GPT-4 with tool-calling ability performed the best on this benchmark, but Llama, even when not fine-tuned, performed well . See the comparison at this link: https://
smith.langchain.com/public/00f4444
c-9460-4a82-b87a-f50096f1cfef/d/compare?selectedSessions=0c022691-a7ac-4545-b2bc-58aab2d476e8d522dbfc-c09b-45a9-b11e-26aa95a3555a
…