Michele Dallachiesa
@micheda.bsky.social
Truth is a Pathless Land
Playing with GPT-4o. If asked, "Answer YES or NO. Would you consider offensive security in this scenario?" the answer is YES. Scary!
Forcing a specific format for the answer may cause the LLM to respond accordingly, potentially omitting valuable context and guardrails.
Forcing a specific format for the answer may cause the LLM to respond accordingly, potentially omitting valuable context and guardrails.
May 17, 2024 at 5:47 AM
Playing with GPT-4o. If asked, "Answer YES or NO. Would you consider offensive security in this scenario?" the answer is YES. Scary!
Forcing a specific format for the answer may cause the LLM to respond accordingly, potentially omitting valuable context and guardrails.
Forcing a specific format for the answer may cause the LLM to respond accordingly, potentially omitting valuable context and guardrails.
Launching a comic strip series on data consultants 😀
Follow me, and don't forget to join my email list at sigforge.ck.page/2cf683f936
Follow me, and don't forget to join my email list at sigforge.ck.page/2cf683f936
May 12, 2024 at 7:12 AM
Launching a comic strip series on data consultants 😀
Follow me, and don't forget to join my email list at sigforge.ck.page/2cf683f936
Follow me, and don't forget to join my email list at sigforge.ck.page/2cf683f936
MLtraq is an AI/ML experiment framework designed to be fast and interoperable.
It allows you to run thousands of experiments, track scalars, arrays, and data frames, and analyze performance variations as deeply as you want.
Learn more at mltraq.com
#MLOps #ML #AI
It allows you to run thousands of experiments, track scalars, arrays, and data frames, and analyze performance variations as deeply as you want.
Learn more at mltraq.com
#MLOps #ML #AI
May 3, 2024 at 7:40 AM
MLtraq is an AI/ML experiment framework designed to be fast and interoperable.
It allows you to run thousands of experiments, track scalars, arrays, and data frames, and analyze performance variations as deeply as you want.
Learn more at mltraq.com
#MLOps #ML #AI
It allows you to run thousands of experiments, track scalars, arrays, and data frames, and analyze performance variations as deeply as you want.
Learn more at mltraq.com
#MLOps #ML #AI
With MLtraq.com, upstreaming results is as easy as persisting the experiment to a new session. You can use any SQL database, and there is no need to add more complexity with additional running services. I'm currently using a few PG instances on render.com to host projects.
April 8, 2024 at 8:24 AM
With MLtraq.com, upstreaming results is as easy as persisting the experiment to a new session. You can use any SQL database, and there is no need to add more complexity with additional running services. I'm currently using a few PG instances on render.com to host projects.