DEV Community

Sarthak Agrawal profile picture

Sarthak Agrawal

404 bio not found

Location Lucknow, India Joined Joined on  github website

Work

CSE Junior at MIT, Manipal

DPO fixed the collapse and still failed the actual task

DPO fixed the collapse and still failed the actual task

1
Comments
1 min read
A retry decision was enough to close the first factory loop

A retry decision was enough to close the first factory loop

Comments
1 min read
One token system fixed a site that felt like two products

One token system fixed a site that felt like two products

1
Comments 1
1 min read
PostTrainLLM vs Axolotl: Mac specialist workflow or accelerator framework?

PostTrainLLM vs Axolotl: Mac specialist workflow or accelerator framework?

1
Comments
2 min read
PostTrainLLM vs Unsloth: start with the hardware and the decision

PostTrainLLM vs Unsloth: start with the hardware and the decision

1
Comments
2 min read
CodeVetter vs Greptile: repository-aware review or task evidence?

CodeVetter vs Greptile: repository-aware review or task evidence?

1
Comments
2 min read
A small language model specialist should earn one job

A small language model specialist should earn one job

1
Comments
2 min read
CodeVetter vs CodeRabbit: review guidance or verification evidence?

CodeVetter vs CodeRabbit: review guidance or verification evidence?

Comments
2 min read
An agent trajectory dataset needs more than final answers

An agent trajectory dataset needs more than final answers

Comments
2 min read
Meeting notes can stay on the Mac

Meeting notes can stay on the Mac

1
Comments
2 min read
What CodeVetter's public benchmark proves, and what it does not

What CodeVetter's public benchmark proves, and what it does not

2
Comments
2 min read
Evaluate a local LLM for a decision, not a leaderboard

Evaluate a local LLM for a decision, not a leaderboard

1
Comments
2 min read
A Mac AI assistant should verify its actions

A Mac AI assistant should verify its actions

2
Comments
2 min read
What belongs in a coding-agent verification evidence bundle?

What belongs in a coding-agent verification evidence bundle?

1
Comments 2
2 min read
MLX LoRA fine-tuning: the command is the easy part

MLX LoRA fine-tuning: the command is the easy part

1
Comments
2 min read
What makes a Mac voice assistant private?

What makes a Mac voice assistant private?

Comments
2 min read
AI code review and agent verification are not the same

AI code review and agent verification are not the same

2
Comments 3
2 min read
Training your own LLM usually means building a specialist

Training your own LLM usually means building a specialist

Comments
2 min read
A screenshot is not enough for a screen-aware Mac assistant

A screenshot is not enough for a screen-aware Mac assistant

Comments
2 min read
A practical way to verify AI-generated code

A practical way to verify AI-generated code

Comments
2 min read
What on-device should mean for a Mac AI assistant

What on-device should mean for a Mac AI assistant

1
Comments
2 min read
Fine-tuning an LLM on a Mac is easy to start and hard to evaluate

Fine-tuning an LLM on a Mac is easy to start and hard to evaluate

Comments 1
3 min read
A coding agent changed the code. What proves the task is done?

A coding agent changed the code. What proves the task is done?

Comments 1
3 min read
loading...