Skip to content

DEV Community

Suman Nath

I learn ML by building it, breaking it, and writing down what broke. Currently fine-tuning LLMs (full FT, LoRA, QLoRA) — the walls tutorials skip.

Bengaluru, India Joined on Jun 21, 2026 https://nathlabs.com/

Suman Nath

Jun 29

A Better LLM Judge? The Rubric Made My Small Model Worse

#machinelearning #llm #python #ai

5 min read

Want to connect with Suman Nath?

Create an account to connect with Suman Nath. You can also sign in below to proceed if you already have an account.

Already have an account? Sign in

Suman Nath

Jun 29

LLM-as-a-Judge: I Built One From Scratch, Then Checked It Against Humans

#machinelearning #llm #python #ai

4 min read

Suman Nath

Jun 26

Breaking down the accuracy number: Building an LLM Eval Harness From Scratch

#machinelearning #llm #python #ai

4 min read

Suman Nath

Jun 21

If a 270M Model Already Worked, Why Did I Fine-Tune a 7B One?

#machinelearning #llm #mlops #ai

3 min read

Suman Nath

Jun 21

QLoRA: Fine-Tuning a 7B Model on a 16GB GPU (It Shrank to 5.4GB in Front of Me)

#machinelearning #llm #python #ai

3 min read

Suman Nath

Jun 21

LoRA: I Trained <1% of a 1.5B Model and Matched a Full Fine-Tune

#machinelearning #llm #python #ai

3 min read

Suman Nath

Jun 21

I Fine-Tuned a 270M Model on My Laptop (Full Fine-Tuning, From Scratch)

#machinelearning #llm #python #ai

2 min read

loading...