Trending

Optimizing Tool Selection for LLM Workflows with Differentiable Programming

The Right Way to Embed an LLM in a Group Chat

Baba Is Eval

What I learned building an AI coding agent for a year

Impact of PCIe 5.0 Bandwidth on GPU Content Creation and LLM Performance

Large Language Models Are Improving Exponentially

Everything around LLMs is still magical and wishful thinking

Gremllm

ChatGPT creates phisher's paradise by serving the wrong URLs for major companies

Is Codestral 22B still the best open LLM for local coding on 32–64 GB VRAM?

How do I build an agent as a standalone product?

When Should We Expect Affordable Hardware That Will Run Large LLMs With Usable Speed?

Llama-4-Maverick 402B on a oneplus 13

Karpathy's analogy about LLMs at a YC Talk [D]

Powerful 4B Nemotron based finetune

AI agent memory that doesn't suck - a practical guide

Google finds LLMs can hide secret information and reasoning in their outputs, and we may soon lose the ability to monitor their thoughts

Open source tool for generating training datasets from text files and pdf for fine-tuning language models.

Tencent

Save item

Hunyuan-A13B-Instruct

Z.ai & THUKEG

Save item

GLM-4.1V-9B-Thinking

Task: image-text-to-text

TNG Technology Consulting GmbH

Save item

DeepSeek-TNG-R1T2-Chimera

Task: text-generation

SK Telecom

Save item

A.X-4.0

Task: text-generation

Agentica

Save item

DeepSWE-Preview

Task: text-generation

BAIDU

Save item

ERNIE-4.5-21B-A3B-PT

Task: text-generation

SK Telecom

Save item

A.X-4.0-Light

Task: text-generation

K intelligence Mi:dm

Save item

Midm-2.0-Base-Instruct

Task: text-generation

BAIDU

Save item

ERNIE-4.5-0.3B-PT

Task: text-generation

transitive-bullshit

Save item

agentic

AI agent stdlib that works with any LLM and TypeScript AI SDK.

microsoft

Save item

unilm

Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities

neuml

Save item

txtai

💡 All-in-one open-source embeddings database for semantic search, LLM orchestration and language model workflows

Giskard-AI

Save item

giskard

🐢 Open-Source Evaluation & Testing for AI & LLM systems

BlinkDL

Save item

RWKV-LM

RWKV (pronounced RwaKuv) is an RNN with great LLM performance, which can also be directly trained like a GPT transformer (parallelizable). We are at RWKV-7 "Goose". So it's combining the best of RNN and transformer - great performance, linear time, constant space (no kv-cache), fast training, infinite ctx_len, and free sentence embedding.