Manish Chablani – Medium

Manish Chablani

Manish Chablani

Aligning LLMs with Direct Preference Optimization (DPO)— background, overview, intuition and paper…

Direct Preference Optimization (DPO) is a stable, performant, and computationally lightweight, technique for aligning LLM’s with a simple…

3 min readFeb 9, 2024

--

Aligning LLMs with Direct Preference Optimization (DPO)— background, overview, intuition and paper…

--

Manish Chablani

ASR ML Systems: Overview and latest model architectures: Transducers, TDT

Traditional ASR pipeline:

8 min readFeb 8, 2024

--

ASR ML Systems: Overview and latest model architectures: Transducers, TDT

--

Manish Chablani

Large Language Models: Collection of papers, architectures and ideas across different LLM’s: GPT…

Here we summarize key insights from following:

5 min readFeb 3, 2024

--

Large Language Models: Collection of papers, architectures and ideas across different LLM’s: GPT…

--

Manish Chablani

Large Language Models: Collection of papers, architectures and ideas across different LLM’s: PART…

OLMo: https://arxiv.org/abs/2402.00838

4 min readFeb 3, 2024

--

Large Language Models: Collection of papers, architectures and ideas across different LLM’s: PART…

--

Manish Chablani

Vision Transformer (ViT) — AN IMAGE IS WORTH 16X16 WORDS: TRANSFORMERS FOR IMAGE RECOGNITION AT…

Paper Summary: Vision Transformer (ViT)

5 min readFeb 2, 2024

--

Vision Transformer (ViT) — AN IMAGE IS WORTH 16X16 WORDS: TRANSFORMERS FOR IMAGE RECOGNITION AT…

--

Manish Chablani

Recommendation system / Search / IR evaluation metrics

Metrics in Recommendation Systems / Retrival / Ranking:

3 min readDec 28, 2023

--

Recommendation system / Search / IR evaluation metrics

--

Manish Chablani

Embedding-based Retrieval in Facebook Search

Paper: https://arxiv.org/abs/2006.11632

16 min readDec 14, 2023

--

Embedding-based Retrieval in Facebook Search

--

Manish Chablani

Measuring feature importance, removing correlated features

Linear model like linear regression or logistic regression: Identify the coefficients ( β ) in the linear regression equation for each…

6 min readDec 13, 2023

--

Measuring feature importance, removing correlated features

--

Manish Chablani

GPT and other LLM’s: decoder only v/s encoder-decoder models?

Pre LLM, during the times of seq2seq models, encoder-decoder architectures were popular for Q&A, language translation and summarization…

1 min readDec 9, 2023

--

1

--

1

Manish Chablani

Graph Neural Nets Explained: Summary of different graph embedding methods.

Node2Vec :

13 min readDec 5, 2023

--

Graph Neural Nets Explained: Summary of different graph embedding methods.

--

Manish Chablani

Manish Chablani

Head of AI @EightSleep , Marathoner. (Past: AI in healthcare @curaiHQ , DL for self driving cars @cruise , ML @Uber , Early engineer @MicrosoftAzure cloud

Following

Help
Status
About
Careers
Press
Blog
Privacy
Terms
Text to speech
Teams