2 87 37

Pham Van Linh

phamvanlinh143

AI & ML interests

OCR, AI, DL

Recent Activity

liked a model 10 days ago

microsoft/Phi-3.5-vision-instruct

liked a model 12 days ago

zai-org/GLM-OCR

liked a model 12 days ago

Qwen/Qwen3-VL-32B-Instruct

View all activity

Organizations

None yet

upvoted an article 18 days ago

Article

Illustrating Reinforcement Learning from Human Feedback (RLHF)

Dec 9, 2022

•

411

upvoted an article 19 days ago

Article

Building a Fast Multilingual OCR Model with Synthetic Data

23 days ago

•

upvoted an article 20 days ago

Article

Understanding Gemma 3n: How MatFormer Gives You Many Models in One

Jun 26, 2025

•

upvoted an article about 1 month ago

Article

Welcome Gemma 4: Frontier multimodal intelligence on device

Apr 2

•

890

upvoted 5 articles 2 months ago

Article

KV Cache from scratch in nanoVLM

Jun 4, 2025

•

119

Article

Unlocking Longer Generation with Key-Value Cache Quantization

May 16, 2024

•

Article

GGML and llama.cpp join HF to ensure the long-term progress of Local AI

Feb 20

•

505

Article

Continuous batching from first principles

Nov 25, 2025

•

378

Article

Mixture of Experts (MoEs) in Transformers

Feb 26

•

159

upvoted 3 articles 3 months ago

Article

2. Attention Optimizations: From Standard Attention to FlashAttention

Feb 9

•

Article

VLM-OCR Recipes on GPU Infrastructure

Jan 15

•

Article

My Journey Into Vision Models

Apr 12, 2025

•

upvoted 3 papers 3 months ago

Reinforcement Learning via Self-Distillation

Paper • 2601.20802 • Published Jan 28 • 44

Efficient Memory Management for Large Language Model Serving with PagedAttention

Paper • 2309.06180 • Published Sep 12, 2023 • 54

SmolDocling: An ultra-compact vision-language model for end-to-end multi-modal document conversion

Paper • 2503.11576 • Published Mar 14, 2025 • 157

upvoted 2 articles 3 months ago

Article

Performant local mixture-of-experts CPU inference with GPU acceleration in llama.cpp

Jan 30

•

Article

LightOnOCR-2-1B: a lightweight high-performance end-to-end OCR model family

Jan 19

•

upvoted 2 articles 4 months ago

Article

The Optimal Architecture for Small Language Models

Dec 26, 2025

•

120

Article

Tokenization in Transformers v5: Simpler, Clearer, and More Modular

Dec 18, 2025

•

124

upvoted an article 5 months ago

Article

Shrinking Giants: The Quantization Mathematics Making LLMs Accessible

May 3, 2025

•

Pham Van Linh

AI & ML interests

Recent Activity

Organizations

phamvanlinh143's activity

Illustrating Reinforcement Learning from Human Feedback (RLHF)

Building a Fast Multilingual OCR Model with Synthetic Data

Understanding Gemma 3n: How MatFormer Gives You Many Models in One

Welcome Gemma 4: Frontier multimodal intelligence on device

KV Cache from scratch in nanoVLM

Unlocking Longer Generation with Key-Value Cache Quantization

GGML and llama.cpp join HF to ensure the long-term progress of Local AI

Continuous batching from first principles

Mixture of Experts (MoEs) in Transformers

2. Attention Optimizations: From Standard Attention to FlashAttention

VLM-OCR Recipes on GPU Infrastructure

My Journey Into Vision Models

Performant local mixture-of-experts CPU inference with GPU acceleration in llama.cpp

LightOnOCR-2-1B: a lightweight high-performance end-to-end OCR model family

The Optimal Architecture for Small Language Models

Tokenization in Transformers v5: Simpler, Clearer, and More Modular

Shrinking Giants: The Quantization Mathematics Making LLMs Accessible