Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Tiny Recursive Models: A tiny 7M parameter model that recursively refines its answer beats LLMs 100x larger on hard puzzles like ARC-AGI We independently reproduced the paper, corroborated results, and released the weights + API access for those looking to benchmark it 🔍

alphaXiv

32,147 subscribers

52,395 views • 7 months ago •via X (Twitter)

Education Science & Technology

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

starting the week with a true groundbreaking work 💥 Large Language Diffusion Models the first billion-parameter scale diffusion model competitive with its pairs (8B model comparable to LLaMA 3 8B) it gets rid of the michael scott syndrome on existing LLMs

starting the week with a true groundbreaking work 💥 Large Language Diffusion Models the first billion-parameter scale diffusion model competitive with its pairs (8B model comparable to LLaMA 3 8B) it gets rid of the michael scott syndrome on existing LLMs

apolinario 🌐

11,833 views • 1 year ago

From a single image to clean, detailed 3D! Hitem3D just released benchmark results for its v1.5_1536 Pro model -and its fantastic! You clearly see the advantage of the new model: ⬅️ Left v1.5_1536 Pro , right regular 1536p➡️ Here’s what stood out:

From a single image to clean, detailed 3D! Hitem3D just released benchmark results for its v1.5_1536 Pro model -and its fantastic! You clearly see the advantage of the new model: ⬅️ Left v1.5_1536 Pro , right regular 1536p➡️ Here’s what stood out:

Chubby♨️

37,355 views • 7 months ago

STARBOY perceives its environment through a camera, microphone, temperature sensor, and accelerometer. We trained multiple tiny image models that run locally on the device, letting it recognize human faces and hand gestures.

STARBOY perceives its environment through a camera, microphone, temperature sensor, and accelerometer. We trained multiple tiny image models that run locally on the device, letting it recognize human faces and hand gestures.

Daniel Kuntz

114,560 views • 3 months ago

We just shipped support for tool use and JSON mode for DeepSeek R1 Distil-Llama 70B on Groq Inc. 🛠️ The coolest part is that our API now has a `reasoning_format` parameter that lets you control how the model outputs its thought process.

We just shipped support for tool use and JSON mode for DeepSeek R1 Distil-Llama 70B on Groq Inc. 🛠️ The coolest part is that our API now has a `reasoning_format` parameter that lets you control how the model outputs its thought process.

Hatice Ozen

43,225 views • 1 year ago

its so nice to see that leeknow is just like a stray kitten around the kids...just sitting at a random corner looking tiny and adorable 🥹🥹

its so nice to see that leeknow is just like a stray kitten around the kids...just sitting at a random corner looking tiny and adorable 🥹🥹

adnine

30,552 views • 10 months ago

LongWriter Unleashing 10,000+ Word Generation from Long Context LLMs discuss: Current long context large language models (LLMs) can process inputs up to 100,000 tokens, yet struggle to generate outputs exceeding even a modest length of 2,000 words. Through controlled experiments, we find that the model's effective generation length is inherently bounded by the sample it has seen during supervised fine-tuning (SFT). In other words, their output limitation is due to the scarcity of long-output examples in existing SFT datasets. To address this, we introduce AgentWrite, an agent-based pipeline that decomposes ultra-long generation tasks into subtasks, enabling off-the-shelf LLMs to generate coherent outputs exceeding 20,000 words. Leveraging AgentWrite, we construct LongWriter-6k, a dataset containing 6,000 SFT data with output lengths ranging from 2k to 32k words. By incorporating this dataset into model training, we successfully scale the output length of existing models to over 10,000 words while maintaining output quality. We also develop LongBench-Write, a comprehensive benchmark for evaluating ultra-long generation capabilities. Our 9B parameter model, further improved through DPO, achieves state-of-the-art performance on this benchmark, surpassing even much larger proprietary models. In general, our work demonstrates that existing long context LLM already possesses the potential for a larger output window--all you need is data with extended output during model alignment to unlock this capability.

LongWriter Unleashing 10,000+ Word Generation from Long Context LLMs discuss: Current long context large language models (LLMs) can process inputs up to 100,000 tokens, yet struggle to generate outputs exceeding even a modest length of 2,000 words. Through controlled experiments, we find that the model's effective generation length is inherently bounded by the sample it has seen during supervised fine-tuning (SFT). In other words, their output limitation is due to the scarcity of long-output examples in existing SFT datasets. To address this, we introduce AgentWrite, an agent-based pipeline that decomposes ultra-long generation tasks into subtasks, enabling off-the-shelf LLMs to generate coherent outputs exceeding 20,000 words. Leveraging AgentWrite, we construct LongWriter-6k, a dataset containing 6,000 SFT data with output lengths ranging from 2k to 32k words. By incorporating this dataset into model training, we successfully scale the output length of existing models to over 10,000 words while maintaining output quality. We also develop LongBench-Write, a comprehensive benchmark for evaluating ultra-long generation capabilities. Our 9B parameter model, further improved through DPO, achieves state-of-the-art performance on this benchmark, surpassing even much larger proprietary models. In general, our work demonstrates that existing long context LLM already possesses the potential for a larger output window--all you need is data with extended output during model alignment to unlock this capability.

AK

50,969 views • 1 year ago

Microsoft made 100B parameter models run on a single CPU. bitnet.cpp: The official inference framework for 1-bit LLMs. The math behind 1-bit LLMs is what makes them revolutionary. Traditional LLMs use 16-bit floating point weights. Every parameter is a number like 0.0023847 or -1.4729. When you run inference, you multiply these floats together. Billions of times. That's why you need GPUs, they're optimized for floating point matrix multiplication. BitNet b1.58 uses ternary weights: {-1, 0, 1}. That's not a simplification. That's a fundamental change in the math. When your weights are only -1, 0, or 1: → Multiply by 1 = keep the value → Multiply by -1 = flip the sign → Multiply by 0 = skip entirely Matrix multiplication becomes addition and subtraction. No floating point operations. No GPU required. This is why bitnet.cpp achieves: → 2.37x to 6.17x speedup on x86 CPUs → 1.37x to 5.07x speedup on ARM CPUs → 71.9% to 82.2% energy reduction on x86 → 55.4% to 70.0% energy reduction on ARM The speedups scale with model size. Larger models see bigger gains because there are more operations to simplify. A 100B parameter model running at human reading speed (5-7 tokens/second) on a single CPU. That's not optimization. That's a different paradigm. Why 1.58 bits? Because log₂(3) ≈ 1.58. Three possible values = 1.58 bits of information per weight. The key insight: These models aren't quantized after training. They're trained from scratch with ternary weights. The model learns to work within the constraint. No precision loss. No quality tradeoff.

Microsoft made 100B parameter models run on a single CPU. bitnet.cpp: The official inference framework for 1-bit LLMs. The math behind 1-bit LLMs is what makes them revolutionary. Traditional LLMs use 16-bit floating point weights. Every parameter is a number like 0.0023847 or -1.4729. When you run inference, you multiply these floats together. Billions of times. That's why you need GPUs, they're optimized for floating point matrix multiplication. BitNet b1.58 uses ternary weights: {-1, 0, 1}. That's not a simplification. That's a fundamental change in the math. When your weights are only -1, 0, or 1: → Multiply by 1 = keep the value → Multiply by -1 = flip the sign → Multiply by 0 = skip entirely Matrix multiplication becomes addition and subtraction. No floating point operations. No GPU required. This is why bitnet.cpp achieves: → 2.37x to 6.17x speedup on x86 CPUs → 1.37x to 5.07x speedup on ARM CPUs → 71.9% to 82.2% energy reduction on x86 → 55.4% to 70.0% energy reduction on ARM The speedups scale with model size. Larger models see bigger gains because there are more operations to simplify. A 100B parameter model running at human reading speed (5-7 tokens/second) on a single CPU. That's not optimization. That's a different paradigm. Why 1.58 bits? Because log₂(3) ≈ 1.58. Three possible values = 1.58 bits of information per weight. The key insight: These models aren't quantized after training. They're trained from scratch with ternary weights. The model learns to work within the constraint. No precision loss. No quality tradeoff.

Tech with Mak

23,024 views • 2 months ago

We’re releasing model weights for our 8B- parameter Dynamic Byte Latent Transformer, an alternative to traditional tokenization methods with the potential to redefine the standards for language model efficiency and reliability. Learn more about how Dynamic Byte Latent Transformer is paving the way for groundbreaking developments in the field of language modeling. Read the research paper, and download the model and code.

We’re releasing model weights for our 8B- parameter Dynamic Byte Latent Transformer, an alternative to traditional tokenization methods with the potential to redefine the standards for language model efficiency and reliability. Learn more about how Dynamic Byte Latent Transformer is paving the way for groundbreaking developments in the field of language modeling. Read the research paper, and download the model and code.

AI at Meta

195,514 views • 1 year ago

We have released the code and weights for our #CVPR2023 paper "Avatars Grow Legs: Generating Smooth Human Motion from Sparse Tracking Inputs with Diffusion Model"! code: abs: project: The demo is below:

We have released the code and weights for our #CVPR2023 paper "Avatars Grow Legs: Generating Smooth Human Motion from Sparse Tracking Inputs with Diffusion Model"! code: abs: project: The demo is below:

Artsiom Sanakoyeu

35,710 views • 3 years ago

Introducing screengrasp - The new SOTA agent click position model - beating Anthropic Computer Use, OmniParser and Molmo by a wide margin. - Simplified API access to the best click models on the market. 🔁Retweet and DM for free API credits. More details 👇

Introducing screengrasp - The new SOTA agent click position model - beating Anthropic Computer Use, OmniParser and Molmo by a wide margin. - Simplified API access to the best click models on the market. 🔁Retweet and DM for free API credits. More details 👇

Florian S

54,319 views • 1 year ago

The latent space of earlier generative models like GANS can linearly encode concepts of the data. What if the data was model weights? We present weights2weights, a subspace in diffusion weights that behaves as an interpretable latent space over customized diffusion models.

The latent space of earlier generative models like GANS can linearly encode concepts of the data. What if the data was model weights? We present weights2weights, a subspace in diffusion weights that behaves as an interpretable latent space over customized diffusion models.

Amil Dravid

94,226 views • 2 years ago

its the way snghwa only said “that looks good” and hngjng immediately started scanning through the menu looking for the mussels and asked snghwa if he would like it….. to be loved is to be noticed and cared for quietly….. (also snghwa’s tiny nods) 🥺🥺🥺

its the way snghwa only said “that looks good” and hngjng immediately started scanning through the menu looking for the mussels and asked snghwa if he would like it….. to be loved is to be noticed and cared for quietly….. (also snghwa’s tiny nods) 🥺🥺🥺

sankitty

18,028 views • 1 year ago

Today, we are releasing Stable Video Diffusion, our first foundation model for generative AI video based on the image model, Stable Diffusion. As part of this research preview, the code, weights, and research paper are now available. Additionally, today you can sign up for our waitlist to access a new upcoming web experience featuring a Text-To-Video interface. To access the model & sign up for our waitlist, visit our website here:

Today, we are releasing Stable Video Diffusion, our first foundation model for generative AI video based on the image model, Stable Diffusion. As part of this research preview, the code, weights, and research paper are now available. Additionally, today you can sign up for our waitlist to access a new upcoming web experience featuring a Text-To-Video interface. To access the model & sign up for our waitlist, visit our website here:

Stability AI

1,024,415 views • 2 years ago

Reinforcement Learning (RL) has long been the dominant method for fine-tuning, powering many state-of-the-art LLMs. Methods like PPO and GRPO explore in action space. But can we instead explore directly in parameter space? YES we can. We propose a scalable framework for full-parameter fine-tuning using Evolution Strategies (ES). By skipping gradients and optimizing directly in parameter space, ES achieves more accurate, efficient, and stable fine-tuning. Paper: Code:

Reinforcement Learning (RL) has long been the dominant method for fine-tuning, powering many state-of-the-art LLMs. Methods like PPO and GRPO explore in action space. But can we instead explore directly in parameter space? YES we can. We propose a scalable framework for full-parameter fine-tuning using Evolution Strategies (ES). By skipping gradients and optimizing directly in parameter space, ES achieves more accurate, efficient, and stable fine-tuning. Paper: Code:

Yulu Gan

414,886 views • 8 months ago

So tiny...are you actually a human? You said you hope to be stepped on like a tiny ant? That really is a hard to understand request.

So tiny...are you actually a human? You said you hope to be stepped on like a tiny ant? That really is a hard to understand request.

Hugetail🐾🏙️

25,361 views • 12 days ago

🚀 Self-speculation brings 6.75x real speedup for LLM generation with SGLang inference! Same model drafts future tokens in Diffusion mode → then verifies them in AR (causal) mode. One model and one KV cache. Just different attention masks. Thanks to perfect alignment, we get 2× longer acceptance lengths than MTP techniques (Eagle-3, MTP, dFlash). We run 2 forward passes… but the 2× higher acceptance means we break even - and with zero overhead from extra drafter, KV cache, or LM head that comes with MTP - those are not free. Last week we released Nemotron-Labs-Diffusion + Tri-mode LLMs! We did continued pre-training on Ministral-3 models by switching attention patterns (block causal bidirectional). Result: one model that runs AR mode, Diffusion mode, and Self-Speculation. Diffusion mode already shows high benchmark accuracy - excited to see what happens when someone beats left-to-right acceptance! 🔥 Github: Paper: SGLang inference: Try the models on HF:

🚀 Self-speculation brings 6.75x real speedup for LLM generation with SGLang inference! Same model drafts future tokens in Diffusion mode → then verifies them in AR (causal) mode. One model and one KV cache. Just different attention masks. Thanks to perfect alignment, we get 2× longer acceptance lengths than MTP techniques (Eagle-3, MTP, dFlash). We run 2 forward passes… but the 2× higher acceptance means we break even - and with zero overhead from extra drafter, KV cache, or LM head that comes with MTP - those are not free. Last week we released Nemotron-Labs-Diffusion + Tri-mode LLMs! We did continued pre-training on Ministral-3 models by switching attention patterns (block causal bidirectional). Result: one model that runs AR mode, Diffusion mode, and Self-Speculation. Diffusion mode already shows high benchmark accuracy - excited to see what happens when someone beats left-to-right acceptance! 🔥 Github: Paper: SGLang inference: Try the models on HF:

Pavlo Molchanov

65,916 views • 15 days ago

it's crazy what a 1.5B model can do these days! "VibeThinker-1.5B is a 1.5-billion parameter dense language model. With a total training cost of only $7,800 USD, it achieves reasoning performance comparable to larger models like GPT OSS-20B Medium." runs perfectly on device!

it's crazy what a 1.5B model can do these days! "VibeThinker-1.5B is a 1.5-billion parameter dense language model. With a total training cost of only $7,800 USD, it achieves reasoning performance comparable to larger models like GPT OSS-20B Medium." runs perfectly on device!

Maziyar PANAHI

202,224 views • 6 months ago

The Bee Hummingbird (Mellisuga helenae). The smallest bird in the world. Found in Cuba, it weighs less than a penny and its heart beats up to 1,200 times per minute. A tiny, feathered engine.

The Bee Hummingbird (Mellisuga helenae). The smallest bird in the world. Found in Cuba, it weighs less than a penny and its heart beats up to 1,200 times per minute. A tiny, feathered engine.

𝐓𝐡𝐞 𝐅𝐚𝐛𝐮𝐥𝐨𝐮𝐬 𝐖𝐞𝐢𝐫𝐝 𝐓𝐫𝐨𝐭𝐭𝐞𝐫s

278,707 views • 3 months ago

This peanut-sized chinese model just dethroned Gemini at reading documents. It’s called glm-ocr. it’s a tiny 0.9b parameter vision-language model that is about to replace every expensive ocr api you use. → Handles text, tables, formulas, handwriting → Scored 94.62 on OmniDocBench V1.5 → 8 languages → vLLM, SGLang, Ollama And it’s 100% open-source.

This peanut-sized chinese model just dethroned Gemini at reading documents. It’s called glm-ocr. it’s a tiny 0.9b parameter vision-language model that is about to replace every expensive ocr api you use. → Handles text, tables, formulas, handwriting → Scored 94.62 on OmniDocBench V1.5 → 8 languages → vLLM, SGLang, Ollama And it’s 100% open-source.

Alvaro Cintas

37,477 views • 2 months ago

🚨New paper! Generative models are often “miscalibrated”. We calibrate diffusion models, LLMs, and more to meet desired distributional properties. E.g. we finetune protein models to better match the diversity of natural proteins.

🚨New paper! Generative models are often “miscalibrated”. We calibrate diffusion models, LLMs, and more to meet desired distributional properties. E.g. we finetune protein models to better match the diversity of natural proteins.

Brian L Trippe

20,445 views • 8 months ago