Tag

#large-language-models

582 posts114 followers

Explore Hashnode

Alternatives

Nash Hasan5d ago · 7 min read

The Inference Wall Hits Campus IT: Direct Pass@k Optimization and the Economics of Test-Time Compute

Direct Pass@k Optimization is an inference-time reasoning framework that prunes non-viable language model thought trajectories at early sequence prefixes, balancing solution quality against beam diver

blog.edtechniti.com

Ahmed Adawy5d ago · 9 min read

Your LLM Is Slow. The GPU Might Not Be the Problem.

Your LLM Is Slow. The GPU Might Not Be the Problem. A practical approach to tracing queueing, retrieval, prefill, and decoding in production AI systems. A user asks a question. Your application authen

adawy.hashnode.dev

jaberadam20016d ago · 9 min read

RAG Chunking: Why Chunk Size Is the Wrong Knob

You've swept chunk size from 256 to 1024 tokens. You tried 20% overlap, then 50%. You swapped in a semantic splitter. Your eval score moved a point and a half, and the answers that were wrong are most

simplyexplained.hashnode.dev

Nash HasanOct 20 · 7 min read

Deterministic JSON Reaches Claude on Bedrock: Why Constrained Decoding Changes Campus AI Infrastructure

Executive Thesis Constrained decoding is an inference-level mechanism that enforces strict schema compliance by masking invalid tokens at the decoding stage. Amazon Web Services introduced native cons

1 blog.edtechniti.com

N Chandra Prakash ReddyOct 18 · 10 min read

Tiny Models, Massive Impact: Running Distilled LLMs on AWS Lambda with Zero Cold Start

I got the amazing chance to attend AWS Community Day Bengaluru recently on 11 July 2026. The event was full with excellent speakers and fantastic sessions all around, but one presentation totally alte

devopstour.hashnode.dev

SantoshOct 17 · 15 min read

How to Build an Autonomous Blog Writer with CrewAI and Python

Those who have just started writing, and even experienced writers, face this problem on a day-to-day basis. They sit in front of a laptop screen for a couple of hours before rage-quitting over what t

santoshsdevnotes.hashnode.dev

jaberadam2001Oct 11 · 10 min read

Prompt Caching for AI Agents: Stop Breaking Your Own Cache

Here's a quick audit worth running before you read further. Pull the usage object from your last hundred LLM calls and work out what fraction of input tokens came from the cache. def cache_hit_rate(us

simplyexplained.hashnode.dev

jaberadam2001Oct 10 · 10 min read

The Exception Pass: Catching What AI Summaries Delete

TL;DR An AI summary isn't a smaller version of your document. It's a more typical one. The sentences most likely to vanish are the ones that limit, reverse, or doubt something else: conditions, nega

simplyexplained.hashnode.dev

Vikram Kumar ShuklaOct 3 · 5 min read

Cloudflare just shipped models that don't talk back. That's the point.

On October 1, Cloudflare released Clef and Clef-flash — the first models trained by its own Workers AI team. I almost skipped the announcement. Another 27B open-weights model. Another leaderboard. But

vikramkumarshukla.hashnode.dev

Kevin KiruiOct 2 · 6 min read

A Human in the Loop Does Not Automatically Make an AI Workflow Safe

One of the most common phrases in healthcare AI is: “A human is always in the loop.” It sounds reassuring. But a human reviewing an AI output does not automatically make the workflow safe. The impor

2 kevin-kirui-ai.hashnode.dev

Search Hashnode

Explore Hashnode

Trending tags this week

The Inference Wall Hits Campus IT: Direct Pass@k Optimization and the Economics of Test-Time Compute

Your LLM Is Slow. The GPU Might Not Be the Problem.

RAG Chunking: Why Chunk Size Is the Wrong Knob

Deterministic JSON Reaches Claude on Bedrock: Why Constrained Decoding Changes Campus AI Infrastructure

Tiny Models, Massive Impact: Running Distilled LLMs on AWS Lambda with Zero Cold Start

How to Build an Autonomous Blog Writer with CrewAI and Python

Prompt Caching for AI Agents: Stop Breaking Your Own Cache

The Exception Pass: Catching What AI Summaries Delete

Cloudflare just shipped models that don't talk back. That's the point.

A Human in the Loop Does Not Automatically Make an AI Workflow Safe