Posts
All the articles I've posted.
-
Trading MatMuls for SRAM Lookups: A 3-Bit Edge Architecture
By trading heavy FP16 MatMuls for SRAM lookups and 1-bit additions, our custom quantization pipeline squeezes state-of-the-art models down to approx. 3 bits per weight with minimal accuracy loss. Here is how bypassing Tensor Cores could reshape the design of future edge AI chips.
-
An approach to calibrating LLM reasoning effort
In this blog post, we will discuss about Controlling reasoning effort in LLMs ( the gpt-oss-style ) and Calibrating LLM Reasoning effort via Label-Free Alignment.
-
Stop Using Embeddings for Everything in RAG
Why deterministic query translation should often come before embeddings in enterprise RAG systems, and how to combine both in a hybrid approach.