Search the archive

lines of innovation. Get the learning kit

JOY CAPTION (HUGGING FACE HUB)

An open source VLM designed specifically for image upcaptioning.

INTRODUCING RPBENCH-AUTO

Boson AI has launched RPBench-Auto, an automated evaluation pipeline to benchmark LLMs' role-playing capabilities inspired by ArenaHard and Alpaca Eval.

LIGHTWEIGHT CHAMP: NVIDIA RELEASES SMALL LANGUAGE MODEL WITH STATE-OF-THE-ART ACCURACY

NVIDIA's Mistral-NeMo-Minitron 8B, a compressed version of the 12B model, achieves state-of-the-art accuracy while being small enough to run on RTX workstations.

agent.

Chances are, you've already spoken to it without even noticing... It can:

Pretty wild, huh? Here, give it a call yourself.

Or TLDR readers can get exclusive access to an even more advanced

version here...

ARTIFACTS AVAILABLE TO ALL

Anthropic has made Artifacts generally available, including on mobile.

CEREBRAS INFERENCE

Cerebras' chipset has massive unified memory. As a result, it can sidestep bandwidth issues and serve models at thousands of tokens per second.

OPENAI AIMS TO RELEASE NEW AI MODEL, ‘STRAWBERRY', IN FALL

OpenAI is planning to release a new AI product called "Strawberry" in the fall. It will feature advanced reasoning capabilities, such as the ability to solve previously unseen math problems, and can perform high-level tasks like developing market strategies.

PHYSICS OF LANGUAGE MODELS: PART 3.3

The 3rd part of a series, this paper explores scaling laws and how many bits are required to store knowledge in a model. The answer seems to be around 2 bits of knowledge per parameter.

FIRE-FLYER AI HPC

DeepSeek has released a paper outlining its hardware-software co-design strategy for DL training.

ANNOUNCING HIGGS LLAMA V2

Boson AI has introduced Higgs-Llama-3-70B-v2, a new model that excels in dialog and comprehension benchmarks like Arena-Hard and AlpacaEval 2.0. The model reduces response regeneration rates by 21.6% and boosts day 1 retention by 5.3% compared to Claude 3.5 Sonnet. Enhanced by an in-house reward model, Higgs Judger, it ties with Google's Gemini 1.5 Pro on performance.

deploy, no developer needed. Get 2 months free

HYBRID MODEL POST TRAINING COOKBOOK

Pre-training hybrid (Mamba style) models is different from pre-training normal Transformers. This post explores how to scale different hyperparameters, data acquisition, and others to get the performance you want.

LLAMA DUO: SYSTEM FOR MODEL FALLBACK

This is a framework that fine-tunes small models to serve as a fallback when closed API models go down. It shows how to migrate from a large to a small model smoothly.

LITSERVE

LitServe is an easy-to-use, flexible serving engine for AI models built on FastAPI. Features like batching, streaming, and GPU autoscaling eliminate the need to rebuild a FastAPI server per model.

LLAVA BITNET (HUGGING FACE HUB)

Llava BitNet is the first ternary (-1, 0, 1) weight model trained on VLM tasks. The model, weights, and scripts are in the process of being fully open sourced. The technical report will be released soon and suggests the model has promising performance.

INFERENCE IS FREE AND INSTANT

While large language models (LLMs) may not significantly improve in reasoning capabilities, their decreasing costs and increasing speeds will make them more useful for repetitive tasks. Although these models may lack true understanding, they can still efficiently handle straightforward tasks.

MEMORY-EFFICIENT LLM TRAINING

Online Subspace Descent is a new optimizer that improves the training of LLMs by enhancing memory efficiency.

QWEN 2 AUDIO (HUGGING FACE HUB)

Qwen has released audio input style models that can reason about music, audio, and sound.

SKETCH2SCENE

Using a number of models in sequence, this group was able to create an amazing model that generates fully playable 3D game scenes based on a single input sketch.

THIS 1MM 'FAN ON A CHIP' COULD PUT ACTIVE COOLING INSIDE ULTRA-THIN GADGETS

xMEMS has introduced the XMC-2400 µCooling chip, a 1mm-tall solid-state fan designed to cool down slim devices like smartphones.

Get 60% off with code TLDRAI

ANTHROPIC REVEALS SYSTEM PROMPTS FOR CLAUDE

Anthropic has added system prompts and updated dates for all models.

SALESFORCE INTRODUCES TEXT-TO-VIDEO GENERATION

Salesforce has introduced xGen-VideoSyn-1, a text-to-video (T2V) model that generates realistic scenes from textual descriptions. The model uses a video variational autoencoder (VidVAE) to compress video data, reducing computational demands, and a Diffusion Transformer (DiT) for improved temporal consistency and generalization.

INFINITE DATASET HUB (HUGGING FACE HUB)

Powered by phi-3-mini, this space uses a rarity prompt to generate data about any topic. It isn't the most accurate, but it is fascinating and powerful.

NEURAL ASSETS

Neural networks can represent and manipulate 3D objects in 2D scenes by conditioning on per object representations. This work may well be the Holy Grail of 3D object disentangling.

TEXT-GUIDED 3D ANIMATIONS

Researchers have introduced a new method called T3M for creating 3D animations guided by text inputs. Unlike previous techniques that relied only on speech, T3M allows for more accurate and customizable animations, making it a valuable tool for virtual reality, gaming, and film production.

BIREFNET

New state of the art for background removal with bireference segmentation.

RB MODULATION

Google has an extremely novel way to personalize diffusion models that outperforms a number of common methods. It is available for PyTorch and with some slight modifications can work with Flux.

IMAGE EDITING WITH VISION LANGUAGE MODELS

FlexEdit is an image editing method that combines Vision Large Language Models (VLLMs) with free-shape masks for more precise edits based on language instructions.

AI COMPANIES ARE PIVOTING FROM CREATING GODS TO BUILDING PRODUCTS

AI companies are struggling to find product-market fit for LLMs, leading to significant investments yet limited commercial success. The five main challenges hindering AI product viability are cost, reliability, privacy concerns, safety and security issues, and user interface limitations. Overcoming these sociotechnical issues is critical for the effective integration and widespread adoption of AI in consumer products.

D-ID LAUNCHES AN AI VIDEO TRANSLATION TOOL THAT INCLUDES VOICE CLONING AND LIP SYNC

D-ID has launched an AI Video Translate feature that clones the speaker's voice and syncs lip movements in translated videos. Supporting 30 languages, it seeks to reduce localization costs for global campaigns. It is available to subscribers, with plans starting at $56 per year. The technology competes with similar offerings from companies like YouTube and Vimeo, as well as numerous AI voice cloning tools.

MY FRIEND, CLAUDE

This author's reliance on Claude, an LLM from Anthropic, for technical writing due to increased work demands highlights LLMs' growing utility in professional contexts. Despite needing expert verification, Claude's assistance has proven cost-effective and underscores a rapidly changing landscape for niche experts facing AI-driven automation. The author reflects on the potential shift in knowledge work as AI tools like Claude become more integrated into routine tasks.

SHELLMATE AI (PRODUCT)

A Mac terminal productivity tool powered by OpenAI.

QUICK FINE-TUNING OF PHI 3.5 (COLAB NOTEBOOK)

Quick fine-tuning script with Unsloth of the new Microsoft models.

VYOND PUSHES AI VIDEO'S ENTERPRISE ERA

Vyond is an AI video creation platform targeted at enterprise use cases.

Cursor, Jasper, and many others. Start integrating today.

MICROSOFT RELEASES POWERFUL NEW PHI-3.5 MODELS, BEATING GOOGLE, OPENAI, AND MORE

Microsoft launched three new models in its Phi series: Phi-3.5-mini-instruct, Phi-3.5-MoE-instruct, and Phi-3.5-vision-instruct, each addressing different AI tasks with impressive benchmark performances. These models are open source under the MIT License and available for developers on Hugging Face. Despite their smaller size compared to some counterparts, the Phi models have achieved near-state-of-the-art results, surpassing competitors like GPT-4o and Llama in certain benchmarks.

WHAT'S REALLY GOING ON IN MACHINE LEARNING

Stephen Wolfram, the creator of Wolfram|Alpha, wrote a great exploration about different phenomena going on in machine learning models. He discovered a set of dynamics about the non-linearities in simple neural networks.

AGI SAFETY AND ALIGNMENT AT GOOGLE DEEPMIND

Google DeepMind's AGI Safety & Alignment team shared a detailed update on their work focused on existential risk from AI. Key areas include amplified oversight, frontier safety, and mechanistic interpretability, with ongoing efforts to refine their approach to technical AGI safety. They highlighted recent achievements, collaborations, and plans to address emerging challenges.

ENERGY-EFFICIENT AUTONOMOUS DRIVING WITH SPIKING NEURAL NETWORKS

Spiking Autonomous Driving (SAD) is the first unified Spiking Neural Network (SNN) aimed at addressing the energy challenges of autonomous driving.

EFFICIENT LANGUAGE MODEL TRAINING

Inheritune is a streamlined method for developing smaller base language models from larger ones by inheriting a few transformer blocks and training on a tiny fraction (0.1%) of the original data. This approach enabled the creation of a 1.5B parameter model using only an A6000 GPU in less than half a day, achieving results that compete well against larger models trained on significantly more data.