- Select Tickets
- Payment
- invoice
- Attendee details
Membership
The Fifth Elephant annual membership
The Fifth Elephant membership is valid for one year - 12 months. The member get the following benefits:
- Participation in all online peer review sessions.
- Access to all recordings from online reviews.
- Priority access to all offline meet-ups and online workshops hosted by The Fifth Elephant during the one year period.
- Access to The Fifth Elephant’s Annual Conference on 18 and 19 July 2025 in Bangalore - in-person and virtually (via live stream).
Corporate Members-only benefits (bulk ticket purchase):
- Transfer of memberships across individuals in the organization.
Memberships can be cancelled within 1 hour of purchase.
₹5100
Sale at this price closes on December 31, 2025
Total ₹0
Cancellation and refund policy
Memberships can be cancelled within 1 hour of purchase
Workshop tickets can be cancelled or transferred upto 24 hours prior to the workshop.
For further queries, please write to us at support@hasgeek.com or call us at +91 7676 33 2020.Tuning a base language model for multi-tasking
Submitted Jun 30, 2023
Abstract
Auquan is an AI startup that serves institutional investors and investment managers with curated news and documents to help them make better investment decisions.
In this presentation, I will discuss our approach for tuning a base language model for multiple tasks, such as determining noise in streaming news feeds, determining relevance, matching news to topics, and curating relevant documents.
I will walk through the process and pitfalls for tuning the language model for a general use case, including the process and metric for determining performance of the tuned model.
Audience
ML Engineers, early stage Data Scientists
Takeaways
- How to tune a base language model for multiple tasks
- Existing libraries for tuning language models
- Best practices and pitfalls for tuning language models
Presentation Outline
- Introduction
- About Auquan and our use case
- Problem description
- Language models for multi-tasking
- How we use language models
- Tuning an LM
- Using tuned models for embedding
- Best practices and pitfalls
- Conclusion/QA
Comments
Hybrid access (members only)
Hosted by
Jump starting better data engineering and AI futures
Supported by
E2E Cloud is India's first AI hyper scaler, a cloud computing platform providing accelerated cloud-based solutions at maximum optimization and lowest pricing
Login to leave a comment
Samik Raychaudhuri
@samikr Submitter
Hello Nischal,
Thanks. Indeed your suggestion is already included in the talk:
"I will walk through the process and pitfalls for tuning the language model for a general use case, including the process and metric for determining performance of the tuned model."
Sure, I can cover this in greater detail. However, as mentioned, the other talk (https://hasgeek.com/fifthelephant/2023/sub/using-lm-and-vector-database-for-large-scale-docum-XziNZdPXcWc1DmSYjhnFPi) is probably a better fit, and can contain the gist of this talk as well.
Please let me know what you think.
Anwesha Sen
@anwesha25 Editor & Promoter
Hello Samik, please drop an email at anwesha@hasgeek.com so I can schedule your rehearsal. You can choose to walk us through both topics, select one, or combine the two. Looking forward!
Nischal HP
@nischalhp Editor
Hello Samik Raychaudhuri,
Thank you for your submission. The outline reads quite well and it is very interesting to see the problem statement and solution in the said space.
It would very valuable to the attendees if you could also add a section which talks about how you measure the success of the model and the impact of it on business decisions, if its already being used in production.
Otherwise, it looks good and we are currently reviewing the talks. We will get back to you with updates shortly.
Samik Raychaudhuri
@samikr Submitter
Hi, I have submitted 2 titles with some overlap. Open to suggestion about which one is more appropriate/interesting for the community. Thanks!