Tweets

2022-11-28T14:23:31+00:00 | 🔗

@rguha @OlorenAI @OlorenChemEng We landed on this approach because MLOps workflow depends heavily on the type of model. E.g. there is https://t.co/0q2H8msyKu for images and hugging https://t.co/X6TREuT0vy for transformers, but we didn’t find much for chemistry…

2022-11-28T14:20:46+00:00 | 🔗

@rguha We are building such a system @OlorenAI leveraging the common interface of @OlorenChemEng to provide a consistent interface for saving/loading/accessing/storing the models. Arbitrary models can be defined as a custom class, which requires a few methods to be implemented.

2022-11-24T16:59:08+00:00 | 🔗

@balajis Counterpoint: He should stay and allow SBF to incriminate himself to get justice?

2022-11-22T17:42:01+00:00 | 🔗

#AI #LLM #GPT4

2022-11-22T17:42:01+00:00 | 🔗

Large language models will be a commodity differentiated by latency and cost. Developer tools will exist but there will be a limited set: fine-tuning and prompt-management. Because of the former, LLMs will be cheap, and because of the latter, there will be a handful of providers

2022-11-22T17:32:04+00:00 | 🔗

@realGeorgeHotz @triplebankshot I would pay for Twitter Blue with this feature (maybe + meaning-based search). That should be enough to pay for the storage/compute required to persist data?

2022-11-19T22:23:29+00:00 | 🔗

@Snowden It’s almost like people only ask questions where the answer is interesting… e.g. 50/50

2022-11-19T04:02:39+00:00 | 🔗

@andrewwhite01 Yes! Would be very interested in your work process :) also… graph rewiring?

2022-11-18T07:05:44+00:00 | 🔗

@Dereklowe @Dereklowe working on my watercolors, made for great reference image! https://t.co/VoHDMYJKT9

2022-11-18T02:49:51+00:00 | 🔗

Does anyone know where “ACS Document 1996” in ChemDraw comes from?