Open source LLMs // are they truly open-source?
A lot of experiments and startup projects based on open-source LLMs, but what about real business?
Open source LLMs are a great initiative, but it is not truly open-source, the weights are closed, it’s just like a binary code, and can be used as is (black box). All risks, bugs, hallucinations, cyber-security issues, etc. are here and can’t be fixed.
Here is the parade of “open-source” LLMs usage examples:
Open-source developers have created thousands of derivatives of models like Llama, including increasingly, mixing models — and they are steadily achieving parity with, or even superiority over closed models on certain metrics (see examples like FinGPT, BioBert, Defog SQLCoder, and Phind).
“A lot of customer are asking themselves: Wait a second, why am I paying for super large model that knows very little about my business? Couldn’t I just use one of these open-source models, and by the way, maybe use a much smaller, open-source model for that (information retrieval) workflow?”
Companies often choose the open source route, he said, when they’re concerned about controlling access to their data, but also when they want more control over the fine-tuning of a model for specialized purposes.
databricks-dolly-15kwas authored by more than 5,000 Databricks employees during March and April of 2023.
We’re certain there’s more out there. Many companies just don’t want to talk publicly about what they’re doing with open-source LLMs or otherwise.
Examples
Same time…
