@mattshumer_ uses expensive LLMs to "teach" smaller LLMs to perform well on a given task.
@wandb's Weave (https://t.co/yzNbkklPUU) pairs well 🍷: code tracking, execution tracing, model versioning/sharing, dataset management, and evaluations!
https://t.co/bVq5XPgzn7
Introducing `llama-405b-to-8b` ✍️
Get the quality of Llama 3.1 405B, at a fraction of the cost and latency.
Give one example of your task, and 405B will teach 8B (~30x cheaper!!) how to do the task perfectly.
And it's open-source: https://t.co/H5590RiFhc
@mattshumer_ uses expensive LLMs to "teach" smaller LLMs to perform well on a given task.
@wandb's Weave (https://t.co/yzNbkklPUU) pairs well 🍷: code tracking, execution tracing, model versioning/sharing, dataset management, and evaluations!
https://t.co/bVq5XPgzn7
Introducing `llama-405b-to-8b` ✍️
Get the quality of Llama 3.1 405B, at a fraction of the cost and latency.
Give one example of your task, and 405B will teach 8B (~30x cheaper!!) how to do the task perfectly.
And it's open-source: https://t.co/H5590RiFhc
@mattshumer_ Wonderful idea @mattshumer_ ! I think @wandb's Weave (https://t.co/yzNbkklPUU) could complement this work nicely (: I made a video showcasing some possible benefits: https://t.co/tfZGislhhc
@mattshumer_ uses expensive LLMs to "teach" smaller LLMs to perform well on a given task.
@wandb's Weave (https://t.co/yzNbkklPUU) pairs well 🍷: code tracking, execution tracing, model versioning/sharing, dataset management, and evaluations!
https://t.co/bVq5XPgzn7
🧵 Weave team is cranking! New eval comparison UI just landed. Get a beautiful high level summary of how your evals stack up, and then drill down to look at actual data in areas of disagreement.
This is a beta feature you can use now. We’d love feedback.
I'm very excited to announce Weave, our new tools to track and evaluate your LLM apps.
Use Weave to:
🍩log and version LLM interactions and surrounding data, from development to production
🍩experiment with prompting techniques, model changes, and parameters
🍩evaluate your models and measure your progress
@anthony_bak@BEBischof@wandb@leland_mcinnes Given a Table, the projector will automatically select either the first column which is list<number>, or if none exists, will autoselect all number columns. Here is an example where we have an image column (which is mapped to the overlay): https://t.co/z9AO3WgDRv
@tylerbeaty_ Nice work @tylerbeaty_ ! Looks sleek and clean - can't wait to try it myself (: I admire your dedication to helping people help become better versions of themselves - keep it up!
We’ve been hard at work building Tables, a new way to organize, understand and improve your data. Today we're opening it up to everyone! Try it here:
https://t.co/QyuLUgS5Nj
When talking to people who haven’t deployed ML models, I keep hearing a lot of misperceptions about ML models in production. Here are a few of them.
(1/6)
@streamlit Awesome work! Giving the community the chance to build components allows us to solve our own problems, and builds up the Streamlit offering way faster than a single team can! Fantastic! (:
@AlecStapp I am pretty sure @profgalloway's headline quote was from his podcast. Specifically the the 18:14 mark of "Get To a Platform" https://t.co/PLXCFY0YVh. He is using a grand statement to elicit the interviewee to also make a strong claim. (although possible he said to wsj as well)
@jasonfried - collaborative authoring
- deep linking to email for easy reference
- tree-based visualization. (replies, reply-all, forwards, change of recipients)... All creates a mess linearly.
- simple, yet appropriately powerful rich text editor
- any method of helping clear/avoid junk