Mudit Bachhawat

ML @google / @iitkgp / I tweet about math, machines and money /

[Mudit Bachhawat @muditbac]

good ideas for optimising context for agents. https://t.co/zjNfRk1lol

[Mudit Bachhawat @muditbac]

for domain specific agentic flows, SLM >> LLM
- reduces cost and latency
- agents need only a narrow subset of an LLM's capabilities with strict output format
https://t.co/Ms35QB8xPK
https://t.co/QhJqzDvLCL

[Jo Kristian Bergum @jobergum]

I’ll add a few other random thoughts

- specialized re-rankers are going away
-more interest in max sim functions for candidate retrieval
- more interest in learned sparse reps
- bm25 is a baseline
- models that can output multiple reps in a single inference (eg m3)
- omnimodels

[Mudit Bachhawat @muditbac]

very interesting take on future with AI https://t.co/kpe9wtzKgA

[Mudit Bachhawat @muditbac]

@henrythe9ths AI Guide

[Mudit Bachhawat @muditbac]

@elonmusk Elon Escobar 😂

[Mudit Bachhawat @muditbac]

good read to understand the spectrum of content discovery system. search, recommendation and everything in between
https://t.co/LWX5EDoWck

[Mudit Bachhawat @muditbac]

finetuning and LoRA causes catastrophic forgetting. context tuning methods like prefix tuning cannot learn new attention patterns. this might be features as it won't loose pretrained knowledge in the model.
https://t.co/YmDiuNtaP2

[Mudit Bachhawat @muditbac]

transferrable skill:
- system design
- first principle thinking
- project management
- unblocking yourself
- relentless focus on impact
- documentation and collaboration
- managing failures
- doing the right thing

beginners focuses on the first, experienced focuses on the latter

[Mudit Bachhawat @muditbac]

in software development, there are two types of skills:

non transferable skills:
- programming language
- coding tools, ide, deployment tools
- framework, build system, etc

[Mudit Bachhawat @muditbac]

Good summary on training new generation of dense retrieval models.
https://t.co/h5tqNljbt8

[Chander Bhatia @ChanderBhatia01]

Stocks don’t hit major peaks and bottoms every month, quarter or year rather once in a decade. For example (data from 2014-2024)-

1- Tata Motors-
2015- 587
2020- 65
2024- 1066 (so far)

2- Kotak Mahindra Bank
2014- 336
2021- 2171
2024- 1544 (so far)

3- Motilal
2014- 22
2018- 394
2020- 120
2024- 690 (so far)

4- ACE
2014- 12
2018- 200
2020- 32
2024- 1695 (so far)

Stocks give you enough opportunities and time to accumulate as well as to sell. If you understand valuation, cyclicality of sectors, quality of management, growth prospects, rerating/derating concepts, you can comfortably earn much more than the benchmarks.

[Mudit Bachhawat @muditbac]

@jobergum Agreed. But this issue will not occur with nDCG because by definition it will require judgements for all the retrieved documents, for each of the source. This problem will occur with metrics like categorical accuracy. Am I missing something here?

[Mudit Bachhawat @muditbac]

@jobergum "the amount of energy needed to refute bullshit is an order of magnitude bigger than what’s needed to produce it"

[Doug Turnbull @softwaredoug]

50+ things AI Engineers should know about search

What am I missing?

https://t.co/64DuBBmTdG

[Mudit Bachhawat @muditbac]

Transforming myself to a memer. One meme at a time.

[Mudit Bachhawat @muditbac]

Super simple implementation of a vectordb. Looking for few folks to build this more as a side project.

Code: https://t.co/FlBDO14oy7
Blog: https://t.co/nOjcrsw2So

[shyamal @shyamalanadkat]

re: finetuning LLMs
0/ you don't always need it. try everything else first.
1/ lot of data + limited compute -> go for PEFT
2/ limited data + lot of compute -> full fine-tuning
3/ limited data + limited compute -> `import openai` ;)

[Mudit Bachhawat @muditbac]

Finetuning + RAG
- retrieve relevant context and finetune using those relevant context to generate final output
- best of both worlds
- needs smart data generation methods

[Mudit Bachhawat @muditbac]

Very practical guide to building with llms. Highly recommended https://t.co/9uz3Mk7smy

[Mudit Bachhawat @muditbac]

*Fine Tuning*
- good for teaching complex instructions. not good for teaching base model knowledge (dont agree to this)
- improves model efficiency and reduces tokens
- knowledge transfer from large model to smaller models