gdiamos

19 ms·

361.
by gdiamos 3y ago
I like blog post title. Introducing Lamini, the LLM Engine for Rapidly Customizing Models Obviously it still takes a huge amount of work to customize a model to be as good as GPT4 or ChatGPT, that’s exactly why we are building Lamini. To gi
362.
by gdiamos 3y ago
This was a frequently asked question among my friends. I’m really curious to see how someone who hasn’t been staring at the docs for weeks would explain it.
363.
by gdiamos 3y ago
Free open source libraries. Paid LLM hosting. 50% cheaper than OpenAI, pay per compute needed to run & create the LLM. Export the weights anytime you want. Enterprise VPC deployments.
364.
by gdiamos 3y ago
Noting that the Github repo includes a data pipeline for instruction fine tunining. What's the difference between this and other data pipelines like Alpaca?
365.
by gdiamos 3y ago
"my demand regarding the ChatGPT policy is that it must be everyone's individual choice to use it. After all, it is my data that I'm putting into it, and I, as a responsible adult, am aware of the data protection issues that
366.
by gdiamos 3y ago
Aren't you worried about ChatGPT downloading all of the IP from Wolfram?
367.
by gdiamos 3y ago
We are going to need better retrieval methods as LLMs augment and generate more content on the internet.
368.
by gdiamos 3y ago
How can I get 128 A100s ?
369.
by gdiamos 3y ago
Not enough people are talking about alignment of these models. ChatGPT is aligned to queries that people submitted to GPT3. If you use it outside of that application it can actually be worse than GPT3.
370.
by gdiamos 3y ago
Can it support hugging face models?
371.
by gdiamos 3y ago
Github link: https://github.com/Barqawiz/IntelliNode
372.
by gdiamos 3y ago
How can you connect the results from the LLM into the planner? Do you build a parser? Do you fine-tune the LLM to adhere to a format that the planner understands?
373.
by gdiamos 3y ago
paywalled :(
374.
by gdiamos 3y ago
How do LLMs help for sequential data modeling for manufacturing and operations?
375.
by gdiamos 7y ago
Press release: https://mlperf.org/press/ Link to the results: https://mlperf.org/inference-results Link to white paper: https://edge.seas.harvard.edu/files/edge/files/mlp
376.
MLPerf Releases Inference Results(mlperf.org)
2 points by gdiamos 7y ago|1 comments
377.
by gdiamos 9y ago
Here we show that over 15 large scale deep neural networks can be trained in mixed IEEE float16 (multiplication) + IEEE float32 (addition) with no loss in accuracy. We are happy to answer questions about this work.
378.
by gdiamos 10y ago
The short answer is yes, training neural networks it is compute bound in the sense that substantially improving computational throughput would enable bigger models to be trained on bigger datasets than is currently possible. The speech rec