正在加载视频...

视频加载失败

返回首页

Molmo by Ai2 - Open source SoTA Multimodal (Vision) Language model, beating Claude 3.5 Sonnet, GPT4V and comparable to GPT4o 🔥 They release four model checkpoints: 1. MolmoE-1B, a mixture of experts model with 1B (active) 7B (total) 2. Molmo-7B-O, most open 7B model 3. Molmo-7B-D, demo model 4....

reach_vb's profile picture

Vaibhav (VB) Srivastav

52,380 subscribers

80,471 次观看 • 1 年前 •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

10 条评论

Vaibhav (VB) Srivastav 的头像
Vaibhav (VB) Srivastav1 年前

Check out their model checkpoints on the Hub:

Gene Sh 的头像
Gene Sh1 年前

@allen_ai Great to see advancements in open source multimodal models!

Carlos Rene | DEGA.org 的头像
Carlos Rene | DEGA.org1 年前

@allen_ai Paul Allen really left an impressive legacy.

CrytpoBro 🇮🇨 的头像
CrytpoBro 🇮🇨1 年前

@allen_ai @ollama

Everyday AI Lab 的头像
Everyday AI Lab1 年前

@EMostaque @allen_ai Heartening to see the how much open source models are progressing.

ppaanngggg 的头像
ppaanngggg1 年前

@allen_ai amazing

Yorkie 的头像
Yorkie1 年前

@allen_ai Nice! New vision model that will need support in llama.cpp!

Alex Reichenbach 的头像
Alex Reichenbach1 年前

@allen_ai Any idea why they didn’t use rlhf/dpo?

iamrobotbear (bk) 的头像
iamrobotbear (bk)1 年前

@allen_ai Wait, @allen_ai Did I miss something, what is the main difference between 7b-o and d? I know the 7B-D is the version on your demo, but in terms of the model or capabilities, I'm a bit confused.

Abdelhamid Mokhtar 的头像
Abdelhamid Mokhtar1 年前

@allen_ai domain name for sale

相关视频

We’re excited to announce the release and open-source of HunyuanImage 3.0 — the largest and most powerful open-source text-to-image model to date, with over 80 billion total parameters, of which 13 billion are activated per token during inference.The effect is completely comparable to the industry’s flagship closed-source model.🚀🚀🚀 HunyuanImage 3.0 originates from our internally developed native multimodal large language model, with fine-tuning and post-training focused on text-to-image generation. This unique foundation gives the model a powerful set of capabilities: ✅Reason with world knowledge ✅Understand complex, thousand-word prompts ✅Generate precise text within images Different from traditional DiT architecture image generation models, HunyuanImage 3.0’s MoE architecture uses a Transfusion-based approach to deeply couple Diffusion and LLM training for a single, powerful system. Built on Hunyuan-A13B, HunyuanImage 3.0 was trained on a massive dataset: 5 billion image-text pairs, video frames, interleaved image-text data, and 6 trillion tokens of text corpora. This hybrid training across multimodal generation, understanding, and LLM capabilities allows the model to seamlessly integrate multiple tasks. Whether you're an illustrator, designer, or creator, this is built to slash your workflow from hours to minutes. HunyuanImage 3.0 can generate intricate text, detailed comics, expressive emojis, and lively, engaging illustrations for educational content. The current release focuses solely on text-to-image generation and future updates will include image-to-image, image editing, multi-turn interaction, and more. 👉🏻Try it now: 🔗GitHub: 🤗Hugging Face:
2:09

We’re excited to announce the release and open-source of HunyuanImage 3.0 — the largest and most powerful open-source text-to-image model to date, with over 80 billion total parameters, of which 13 billion are activated per token during inference.The effect is completely comparable to the industry’s flagship closed-source model.🚀🚀🚀 HunyuanImage 3.0 originates from our internally developed native multimodal large language model, with fine-tuning and post-training focused on text-to-image generation. This unique foundation gives the model a powerful set of capabilities: ✅Reason with world knowledge ✅Understand complex, thousand-word prompts ✅Generate precise text within images Different from traditional DiT architecture image generation models, HunyuanImage 3.0’s MoE architecture uses a Transfusion-based approach to deeply couple Diffusion and LLM training for a single, powerful system. Built on Hunyuan-A13B, HunyuanImage 3.0 was trained on a massive dataset: 5 billion image-text pairs, video frames, interleaved image-text data, and 6 trillion tokens of text corpora. This hybrid training across multimodal generation, understanding, and LLM capabilities allows the model to seamlessly integrate multiple tasks. Whether you're an illustrator, designer, or creator, this is built to slash your workflow from hours to minutes. HunyuanImage 3.0 can generate intricate text, detailed comics, expressive emojis, and lively, engaging illustrations for educational content. The current release focuses solely on text-to-image generation and future updates will include image-to-image, image editing, multi-turn interaction, and more. 👉🏻Try it now: 🔗GitHub: 🤗Hugging Face:

Tencent Hy

412,658 次观看 • 9 个月前

Excited to announce GR00T N1, the world’s first open foundation model for humanoid robots! We are on a mission to democratize Physical AI. The power of general robot brain, in the palm of your hand - with only 2B parameters, N1 learns from the most diverse physical action dataset ever compiled and punches above its weight: - Real humanoid teleoperation data. - Large-scale simulation data: we are open-sourcing 300K+ trajectories! - Neural trajectories: we apply SOTA video generation models to “hallucinate” new synthetic data that features accurate physics in pixels. Using Jensen’s words, “systematically infinite data”! - Latent actions: we develop novel algorithms to extract action tokens from in-the-wild human videos and neural generated videos. GR00T N1 is a single end-to-end neural net, from photons to actions: - Vision-Language Model (System 2) that interprets the physical world through vision and language instructions, enabling robots to reason about their environment and instructions, and plan the right actions. - Diffusion Transformer (System 1) that “renders” smooth and precise motor actions at 120 Hz, executing the latent plan made by System 2. We deploy N1 on GR1 robot, 1X Neo robot, and a large collection of simulation benchmarks. N1 achieves up to +30% boost in diverse manipulation tasks for household and industrial settings. While humanoid robots are the main focus of N1, our model also supports cross-embodiment. We finetune it to work on the $110 HuggingFace LeRobot SO100 robot arm! Open robot brain runs on open hardware. Sounds just right. Let’s solve robotics, together, one token at a time. Links to our Whitepaper, Github repo, HuggingFace model, and open dataset page in the thread: 🧵
2:21

Excited to announce GR00T N1, the world’s first open foundation model for humanoid robots! We are on a mission to democratize Physical AI. The power of general robot brain, in the palm of your hand - with only 2B parameters, N1 learns from the most diverse physical action dataset ever compiled and punches above its weight: - Real humanoid teleoperation data. - Large-scale simulation data: we are open-sourcing 300K+ trajectories! - Neural trajectories: we apply SOTA video generation models to “hallucinate” new synthetic data that features accurate physics in pixels. Using Jensen’s words, “systematically infinite data”! - Latent actions: we develop novel algorithms to extract action tokens from in-the-wild human videos and neural generated videos. GR00T N1 is a single end-to-end neural net, from photons to actions: - Vision-Language Model (System 2) that interprets the physical world through vision and language instructions, enabling robots to reason about their environment and instructions, and plan the right actions. - Diffusion Transformer (System 1) that “renders” smooth and precise motor actions at 120 Hz, executing the latent plan made by System 2. We deploy N1 on GR1 robot, 1X Neo robot, and a large collection of simulation benchmarks. N1 achieves up to +30% boost in diverse manipulation tasks for household and industrial settings. While humanoid robots are the main focus of N1, our model also supports cross-embodiment. We finetune it to work on the $110 HuggingFace LeRobot SO100 robot arm! Open robot brain runs on open hardware. Sounds just right. Let’s solve robotics, together, one token at a time. Links to our Whitepaper, Github repo, HuggingFace model, and open dataset page in the thread: 🧵

Jim Fan

465,968 次观看 • 1 年前

PDF parsing is still painful because LLMs reorder text in complex layouts, break tables across pages, and fail on graphs or images. 💡Testing the new open-source OCRFlux model, and here the results are really good for a change. So OCRFlux is a multimodal, LLM based toolkit for converting PDFs and images into clean, readable, plain Markdown text. Because the underlying VLM is only 3B param, it runs even on a 3090 GPU. The model is available on Hugging Face . The engine that powers the OCRFlux, teaches the model to rebuild every page and then stitch fragments across pages into one clean Markdown file. It bundles one vision language model with 3B parameters that was fine-tuned from Qwen 2.5-VL-3B-Instruct for both page parsing and cross-page merging. OCRFlux reads raw page images and, guided by task prompts, outputs Markdown for each page and merges split elements across pages. The evaluation shows Edit Distance Similarity (EDS) 0.967 and cross‑page table Tree Edit Distance 0.950, so the parser is both accurate and layout aware. How it works while parsing each page - Convert into text with a natural reading order, even in the presence of multi-column layouts, figures, and insets - Support for complicated tables and equations - Automatically removes headers and footers Cross-page table/paragraph merging - Cross-page table merging - Cross-page paragraph merging A compact vision‑language models can beat bigger models once cross‑page context is added. 🧵 1/n Read on 👇
0:44

PDF parsing is still painful because LLMs reorder text in complex layouts, break tables across pages, and fail on graphs or images. 💡Testing the new open-source OCRFlux model, and here the results are really good for a change. So OCRFlux is a multimodal, LLM based toolkit for converting PDFs and images into clean, readable, plain Markdown text. Because the underlying VLM is only 3B param, it runs even on a 3090 GPU. The model is available on Hugging Face . The engine that powers the OCRFlux, teaches the model to rebuild every page and then stitch fragments across pages into one clean Markdown file. It bundles one vision language model with 3B parameters that was fine-tuned from Qwen 2.5-VL-3B-Instruct for both page parsing and cross-page merging. OCRFlux reads raw page images and, guided by task prompts, outputs Markdown for each page and merges split elements across pages. The evaluation shows Edit Distance Similarity (EDS) 0.967 and cross‑page table Tree Edit Distance 0.950, so the parser is both accurate and layout aware. How it works while parsing each page - Convert into text with a natural reading order, even in the presence of multi-column layouts, figures, and insets - Support for complicated tables and equations - Automatically removes headers and footers Cross-page table/paragraph merging - Cross-page table merging - Cross-page paragraph merging A compact vision‑language models can beat bigger models once cross‑page context is added. 🧵 1/n Read on 👇

Rohan Paul

149,292 次观看 • 1 年前

Want to create an avatar from a single image? FlexAvatar is a transformer model that creates full 360°, high-quality, and expressive 3D head avatar from just a single portrait image in minutes. Real-time Demo: FlexAvatar's lightweight architecture allows both animation and rendering in real-time, enabling interactive user experiences. To create a new 3D head avatar, only one image is required, e.g., from a webcam. The final avatar is ready after 2 minutes. Architecture: Under the hood, FlexAvatar adopts a transformer-based encoder-decoder design. The encoder maps the input image onto a latent avatar space, while the decoder produces 3D Gaussian attribute maps by incorporating the animation signal via cross-attention. The model learns all facial animations directly from the data without relying on pre-built 3D face models. This equips the avatars with realistic facial expressions. The internal avatar latent space can be conveniently used to integrate additional observations of a person via fitting. This enables use-cases where more than one image of a person is available, e.g., from a phone scan of the person. We train jointly on 2D monocular videos and multi-view data. However, in monocular videos, the animation signal leaks the target viewpoint, causing the model to produce incomplete 3D heads. We call this phenomenon entanglement of driving signal and target viewpoint. To prevent entanglement, we introduce bias sinks. These are learnable tokens that indicate whether a training sample stems from a monocular or a multi-view dataset. During training, the model learns to produce incomplete 3D heads only when the monocular token is present. During inference, FlexAvatar then always uses the multi-view token for which the model has learned to produce complete 3D heads. This simple design allows to combine the generalizability from monocular data with the quality of multi-view data. FlexAvatar summary: - Input: Single-image, phone scan, or monocular video - Output: Full 360° head avatar - Expressive animations - Real-time rendering and animation - Generalization to any portrait - Create a new avatar in 2 minutes - Use bias sinks to combine 2D and 3D data 🏠 🌍 🎥 Great work by Tobias Kirschstein and Simon Giebenhain!
0:30

Want to create an avatar from a single image? FlexAvatar is a transformer model that creates full 360°, high-quality, and expressive 3D head avatar from just a single portrait image in minutes. Real-time Demo: FlexAvatar's lightweight architecture allows both animation and rendering in real-time, enabling interactive user experiences. To create a new 3D head avatar, only one image is required, e.g., from a webcam. The final avatar is ready after 2 minutes. Architecture: Under the hood, FlexAvatar adopts a transformer-based encoder-decoder design. The encoder maps the input image onto a latent avatar space, while the decoder produces 3D Gaussian attribute maps by incorporating the animation signal via cross-attention. The model learns all facial animations directly from the data without relying on pre-built 3D face models. This equips the avatars with realistic facial expressions. The internal avatar latent space can be conveniently used to integrate additional observations of a person via fitting. This enables use-cases where more than one image of a person is available, e.g., from a phone scan of the person. We train jointly on 2D monocular videos and multi-view data. However, in monocular videos, the animation signal leaks the target viewpoint, causing the model to produce incomplete 3D heads. We call this phenomenon entanglement of driving signal and target viewpoint. To prevent entanglement, we introduce bias sinks. These are learnable tokens that indicate whether a training sample stems from a monocular or a multi-view dataset. During training, the model learns to produce incomplete 3D heads only when the monocular token is present. During inference, FlexAvatar then always uses the multi-view token for which the model has learned to produce complete 3D heads. This simple design allows to combine the generalizability from monocular data with the quality of multi-view data. FlexAvatar summary: - Input: Single-image, phone scan, or monocular video - Output: Full 360° head avatar - Expressive animations - Real-time rendering and animation - Generalization to any portrait - Create a new avatar in 2 minutes - Use bias sinks to combine 2D and 3D data 🏠 🌍 🎥 Great work by Tobias Kirschstein and Simon Giebenhain!

Matthias Niessner

95,918 次观看 • 6 个月前

There is a beautiful story that just happened in AI so let me share it for a lighter tone weekend post among all the doom stories in our AI field this week. It’s a story of people on three continents building and sharing in the open a new small efficient and state-of-the-art AI model. It started a couple of months ago when a new team in the AI scene released their first model from their headquarters in Paris (France): Mistral 7B. Impressive model, small and very strong performances in the benchmarks, better than all previous models of this size. And open source! So you could build on top of it. Lewis in Bern (Switzerland) and Ed (in Lyon, in the South of France) both from the H4 team, a team of researchers in model fine-tuning and alignment were talking about it over a coffee, in one of these gatherings that often happen at Hugging Face to break the distance between people (literal distance as HF is a remote company). What about fine-tuning it using this new DPO method that a research team from Stanford in California just posted on Arxiv, says one? Hey, that’s a great idea, replies the other. We've just build a great code base (with Nathan, Nazneen, Costa, Younes and all the H4 team and TRL community) let's use it! The next day they start diving in the datasets openly shared on the HF hub and stumble upon two interesting large and good quality fine-tuning datasets recently open-sourced by OpenBMB, a Chinese team from Tsinghua: UltraFeedback and UltraChat. A few rounds of training experiments confirm the intuition, the resulting model is super strong, by far the strongest they have ever seen in their benchmarks from Berkeley and Stanford (LMSYS and Alpaca). Join Clementine, the big boss of the open evaluation leaderboard. Her deep dive into the model capabilities confirms the results: impressive performance. But the H4 team also hosts a famous faculty member, Pr. Sasha Rush, Associate Professor at Cornell University in his daytime, hacker at HF in his nighttime. Joining the conversation, he proposes to quickly draft a research paper to organize and share all the details with the community. A few days later, the model, called Zephyr (a wind like Mistral), paper, and all details are shared with the world. Quickly other companies, everywhere in the world starts to use it. LlamaIndex, a famous data framework and community, shares how the model blew their expectations on real-life use-case benchmarks, while researchers and practitioners discuss the paper and work on the Hugging Face hub. All this happened in just a few weeks catalyzed by open access to knowledge, models, research, and datasets released all over the world (Europe, California, China) and by the idea that people can build upon one another work in AI to bring real-world value with efficient and open models. Stories like this are numerous everywhere around us and make me really proud of the AI community and see how we can build amazingly useful things together. [the video is just me reading this Friday post hahah]
4:30

There is a beautiful story that just happened in AI so let me share it for a lighter tone weekend post among all the doom stories in our AI field this week. It’s a story of people on three continents building and sharing in the open a new small efficient and state-of-the-art AI model. It started a couple of months ago when a new team in the AI scene released their first model from their headquarters in Paris (France): Mistral 7B. Impressive model, small and very strong performances in the benchmarks, better than all previous models of this size. And open source! So you could build on top of it. Lewis in Bern (Switzerland) and Ed (in Lyon, in the South of France) both from the H4 team, a team of researchers in model fine-tuning and alignment were talking about it over a coffee, in one of these gatherings that often happen at Hugging Face to break the distance between people (literal distance as HF is a remote company). What about fine-tuning it using this new DPO method that a research team from Stanford in California just posted on Arxiv, says one? Hey, that’s a great idea, replies the other. We've just build a great code base (with Nathan, Nazneen, Costa, Younes and all the H4 team and TRL community) let's use it! The next day they start diving in the datasets openly shared on the HF hub and stumble upon two interesting large and good quality fine-tuning datasets recently open-sourced by OpenBMB, a Chinese team from Tsinghua: UltraFeedback and UltraChat. A few rounds of training experiments confirm the intuition, the resulting model is super strong, by far the strongest they have ever seen in their benchmarks from Berkeley and Stanford (LMSYS and Alpaca). Join Clementine, the big boss of the open evaluation leaderboard. Her deep dive into the model capabilities confirms the results: impressive performance. But the H4 team also hosts a famous faculty member, Pr. Sasha Rush, Associate Professor at Cornell University in his daytime, hacker at HF in his nighttime. Joining the conversation, he proposes to quickly draft a research paper to organize and share all the details with the community. A few days later, the model, called Zephyr (a wind like Mistral), paper, and all details are shared with the world. Quickly other companies, everywhere in the world starts to use it. LlamaIndex, a famous data framework and community, shares how the model blew their expectations on real-life use-case benchmarks, while researchers and practitioners discuss the paper and work on the Hugging Face hub. All this happened in just a few weeks catalyzed by open access to knowledge, models, research, and datasets released all over the world (Europe, California, China) and by the idea that people can build upon one another work in AI to bring real-world value with efficient and open models. Stories like this are numerous everywhere around us and make me really proud of the AI community and see how we can build amazingly useful things together. [the video is just me reading this Friday post hahah]

Thomas Wolf

169,127 次观看 • 2 年前

Google just proved that bigger isn't always better. Their 308M parameter model is outperforming models 2x its size. Google just released 𝗘𝗺𝗯𝗲𝗱𝗱𝗶𝗻𝗴𝗚𝗲𝗺𝗺𝗮, and it's proving that lightweight embedding models can punch way above their weight class. At just 308M parameters (578MB), it's the new state-of-the-art for models under 500M parameters across MTEB multilingual, English, and code benchmarks. But the really impressive part is that it ranks 8th overall on MTEB(Multilingual, v2) - that's 𝟭𝟳 𝗽𝗹𝗮𝗰𝗲𝘀 above the second-best sub-500M model, and it's delivering performance 𝗰𝗼𝗺𝗽𝗮𝗿𝗮𝗯𝗹𝗲 𝘁𝗼 𝗺𝗼𝗱𝗲𝗹𝘀 𝗻𝗲𝗮𝗿𝗹𝘆 𝗱𝗼𝘂𝗯𝗹𝗲 𝗶𝘁𝘀 𝘀𝗶𝘇𝗲. There are three key parts of their training recipe that sets it apart: 𝟭. 𝗘𝗻𝗰𝗼𝗱𝗲𝗿-𝗗𝗲𝗰𝗼𝗱𝗲𝗿 𝗜𝗻𝗶𝘁𝗶𝗮𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻 Instead of starting from a decoder-only Gemma 3 model, they first adapted it to encoder-decoder, then used just the encoder. By basing EmbeddingGemma off an LLM that already has world and language understanding, it gives it a stronger starting point. 𝟮. 𝗧𝗵𝗿𝗲𝗲-𝗟𝗼𝘀𝘀 𝗧𝗿𝗮𝗶𝗻𝗶𝗻𝗴 They combine three different loss functions, instead of just having one: • Contrastive loss (NCE) with in-batch negatives and hardness weighting • Spread-out regularization to ensure embeddings utilize the full space (for quantization and ANN retrieval) • Embedding matching distillation from Gemini Embedding - not just learning from relevance scores, but directly aligning the embedding space with the teacher model 𝟯. 𝗠𝗼𝗱𝗲𝗹 𝗦𝗼𝘂𝗽𝗶𝗻𝗴 Rather than just averaging checkpoints from the same training run, they use optimization techniques to find multiple specialized training mixtures. Each mixture creates an "expert" model in different domains, and averaging all their parameters creates a final model that's actually better than individual models. Extras: • Matryoshka embeddings supporting 768, 512, 256, and 128 dimensions • Quantization-aware training - maintains quality even at int4 precision • 100+ languages from Gemma 3 pretraining • Exceptional performance on low-resource languages (check their XTREME-UP results) Is it the absolute best embedding model? No - Gemini Embedding still leads overall. But that's not really the point. EmbeddingGemma proves you can achieve state-of-the-art performance in a small package that's actually deployable on-device, in low-latency applications, and in resource-constrained environments. This makes good embeddings accessible for use cases that I'm seeing more and more: offline applications, privacy-sensitive deployments, and high-throughput scenarios where inference cost actually matters. Full paper: Shoutout to the EmbeddingGemma team at Google DeepMind for this awesome open source work 💙 and to Daniel Williams for helping me with this video! 🫶
5:14

Google just proved that bigger isn't always better. Their 308M parameter model is outperforming models 2x its size. Google just released 𝗘𝗺𝗯𝗲𝗱𝗱𝗶𝗻𝗴𝗚𝗲𝗺𝗺𝗮, and it's proving that lightweight embedding models can punch way above their weight class. At just 308M parameters (578MB), it's the new state-of-the-art for models under 500M parameters across MTEB multilingual, English, and code benchmarks. But the really impressive part is that it ranks 8th overall on MTEB(Multilingual, v2) - that's 𝟭𝟳 𝗽𝗹𝗮𝗰𝗲𝘀 above the second-best sub-500M model, and it's delivering performance 𝗰𝗼𝗺𝗽𝗮𝗿𝗮𝗯𝗹𝗲 𝘁𝗼 𝗺𝗼𝗱𝗲𝗹𝘀 𝗻𝗲𝗮𝗿𝗹𝘆 𝗱𝗼𝘂𝗯𝗹𝗲 𝗶𝘁𝘀 𝘀𝗶𝘇𝗲. There are three key parts of their training recipe that sets it apart: 𝟭. 𝗘𝗻𝗰𝗼𝗱𝗲𝗿-𝗗𝗲𝗰𝗼𝗱𝗲𝗿 𝗜𝗻𝗶𝘁𝗶𝗮𝗹𝗶𝘇𝗮𝘁𝗶𝗼𝗻 Instead of starting from a decoder-only Gemma 3 model, they first adapted it to encoder-decoder, then used just the encoder. By basing EmbeddingGemma off an LLM that already has world and language understanding, it gives it a stronger starting point. 𝟮. 𝗧𝗵𝗿𝗲𝗲-𝗟𝗼𝘀𝘀 𝗧𝗿𝗮𝗶𝗻𝗶𝗻𝗴 They combine three different loss functions, instead of just having one: • Contrastive loss (NCE) with in-batch negatives and hardness weighting • Spread-out regularization to ensure embeddings utilize the full space (for quantization and ANN retrieval) • Embedding matching distillation from Gemini Embedding - not just learning from relevance scores, but directly aligning the embedding space with the teacher model 𝟯. 𝗠𝗼𝗱𝗲𝗹 𝗦𝗼𝘂𝗽𝗶𝗻𝗴 Rather than just averaging checkpoints from the same training run, they use optimization techniques to find multiple specialized training mixtures. Each mixture creates an "expert" model in different domains, and averaging all their parameters creates a final model that's actually better than individual models. Extras: • Matryoshka embeddings supporting 768, 512, 256, and 128 dimensions • Quantization-aware training - maintains quality even at int4 precision • 100+ languages from Gemma 3 pretraining • Exceptional performance on low-resource languages (check their XTREME-UP results) Is it the absolute best embedding model? No - Gemini Embedding still leads overall. But that's not really the point. EmbeddingGemma proves you can achieve state-of-the-art performance in a small package that's actually deployable on-device, in low-latency applications, and in resource-constrained environments. This makes good embeddings accessible for use cases that I'm seeing more and more: offline applications, privacy-sensitive deployments, and high-throughput scenarios where inference cost actually matters. Full paper: Shoutout to the EmbeddingGemma team at Google DeepMind for this awesome open source work 💙 and to Daniel Williams for helping me with this video! 🫶

Victoria Slocum

21,592 次观看 • 7 个月前

Reinforcement Learning from Human Feedback (RLHF) is gaining traction. This field aims to make AI more responsible by including human values and preferences. In this video, Nathan Lambert, a research scientist and RLHF team lead at Hugging Face explores its inner workings, applications and industry impact. RLHF has gained the spotlight in recent years. The growth of language models like Anthropic’s Claude and OpenAI's ChatGPT have increased interest in human-feedback integration. "There are some rumors that Open AI had two teams; one was doing RLHF and the other instruction fine-tuning. And the RLHF team kept getting more and more performance." Understanding RLHF The RLHF process has three main steps: Pre-training: Much like with GPT models, the journey starts with pre-training on a large corpus of data. This can range from text data, web scrapes, to specialized datasets. Reward Modeling: This is the RLHF counterpart of supervised fine-tuning in large language models. This stage involves creating a reward model that resonates with human values and preferences. RL Optimization: This stage parallels reward modeling and reinforcement learning in traditional AI models. The AI system fine-tunes itself based on the reward model, employing reinforcement learning algorithms for that extra layer of optimization. The Data Challenge Data collection and curation in RLHF closely resemble the challenges you'd encounter in large language model training. Datasets from organizations like OpenAI can serve as a useful foundation. However, the need for high-quality, task-specific data cannot be overstated. Implementing RLHF: A Practical Guide If you’re someone who loves getting hands-on with AI libraries like Hugging Face, implementing RLHF is right way to do. It’s essential to understand its limitations. Think about model stability, over-optimization, and exploration strategies, much like you would when prompt engineering. Ongoing Research and Next Steps While he suggests that some basics figured out, there are layers of complexity that still need to be unraveled: 1. New Benchmarks: How do we measure the effectiveness of RLHF? 2. Preference Modeling: How can the model be made to understand human preferences better? 3. Interpreting RLHF: Much like explainability in traditional models, how do we make RLHF more interpretable? 4. System-Wide Evaluation: Going beyond individual performance, how does RLHF affect an entire system? The Transformative Power of RLHF Whether you're an AI developer, a business analyst, or a marketer, RLHF promises to revolutionize your domain. Imagine customer service chatbots that understand human emotions better, or content generators that align more closely with human values. RLHF is an emerging field that focuses on enhancing machine learning models through human feedback. While it tackles important issues like bias and ethics, its broader goal is to improve system performance across various applications. Whether you're deeply invested in the ethics of AI or simply curious about advancements in machine learning, RLHF offers valuable insights. If you're interested in the next wave of AI development, this area is definitely one to watch.
2:00

Reinforcement Learning from Human Feedback (RLHF) is gaining traction. This field aims to make AI more responsible by including human values and preferences. In this video, Nathan Lambert, a research scientist and RLHF team lead at Hugging Face explores its inner workings, applications and industry impact. RLHF has gained the spotlight in recent years. The growth of language models like Anthropic’s Claude and OpenAI's ChatGPT have increased interest in human-feedback integration. "There are some rumors that Open AI had two teams; one was doing RLHF and the other instruction fine-tuning. And the RLHF team kept getting more and more performance." Understanding RLHF The RLHF process has three main steps: Pre-training: Much like with GPT models, the journey starts with pre-training on a large corpus of data. This can range from text data, web scrapes, to specialized datasets. Reward Modeling: This is the RLHF counterpart of supervised fine-tuning in large language models. This stage involves creating a reward model that resonates with human values and preferences. RL Optimization: This stage parallels reward modeling and reinforcement learning in traditional AI models. The AI system fine-tunes itself based on the reward model, employing reinforcement learning algorithms for that extra layer of optimization. The Data Challenge Data collection and curation in RLHF closely resemble the challenges you'd encounter in large language model training. Datasets from organizations like OpenAI can serve as a useful foundation. However, the need for high-quality, task-specific data cannot be overstated. Implementing RLHF: A Practical Guide If you’re someone who loves getting hands-on with AI libraries like Hugging Face, implementing RLHF is right way to do. It’s essential to understand its limitations. Think about model stability, over-optimization, and exploration strategies, much like you would when prompt engineering. Ongoing Research and Next Steps While he suggests that some basics figured out, there are layers of complexity that still need to be unraveled: 1. New Benchmarks: How do we measure the effectiveness of RLHF? 2. Preference Modeling: How can the model be made to understand human preferences better? 3. Interpreting RLHF: Much like explainability in traditional models, how do we make RLHF more interpretable? 4. System-Wide Evaluation: Going beyond individual performance, how does RLHF affect an entire system? The Transformative Power of RLHF Whether you're an AI developer, a business analyst, or a marketer, RLHF promises to revolutionize your domain. Imagine customer service chatbots that understand human emotions better, or content generators that align more closely with human values. RLHF is an emerging field that focuses on enhancing machine learning models through human feedback. While it tackles important issues like bias and ethics, its broader goal is to improve system performance across various applications. Whether you're deeply invested in the ethics of AI or simply curious about advancements in machine learning, RLHF offers valuable insights. If you're interested in the next wave of AI development, this area is definitely one to watch.

Muratcan Koylan

27,005 次观看 • 2 年前

Alibaba just dropped Qwen3.5-397B-A17B and there's a lot to unpack. 397B params, 17B active per forward pass. Sparse MoE done right. But the real story isn't the size—it's the architecture choices. The MoE Design Most MoE models feel like bolt-ons. Qwen 3.5's sparse activation is native—only 4.3% of parameters fire per token. That's how you get trillion-parameter-class performance without trillion-parameter inference costs. The 0.8 RMB/million tokens pricing isn't subsidized; it's structurally earned. Native Multimodal, Not Glued-On This is a vision-language model from the ground up. Heterogeneous architecture—separate processing pipelines for text, image, video that fuse early. Not a vision encoder slapped onto an LLM. The result: 90.8 on OmniDocBench, 79.0 on MMMU-Pro. Document understanding and visual reasoning without the usual brittleness. The Context Window Reality Qwen3.5-Plus (the hosted version) ships with 1M tokens by default. That's not a marketing number—they're actually positioning it for long-document workflows. With built-in adaptive tool use, it's clearly aimed at agentic automation, not just chat. What Actually Impressed Me • FP8 native pipeline: ~50% activation memory reduction • Async RL framework for continuous refinement—training and inference workloads separated • 201 languages (up from 119), 250k vocab for better low-resource encoding • Apache 2.0 license. Full weights on HuggingFace and ModelScope. The Benchmark Context 76.4 on SWE-bench Verified puts it in the range where it can handle real debugging workflows. 72.9 on BFCL v4 for agentic tool use. 88.4 on GPQA Diamond. These aren't SOTA in isolation, but the breadth is unusual—strong across reasoning, coding, multimodal, and agentic tasks. The Honest Caveat I haven't stress-tested the 1M context for needle-in-haystack retrieval yet. And "native multimodal" claims need real-world torture testing—PDFs with tables, charts, mixed layouts. Benchmarks are benchmarks. Bottom Line This isn't just another model release. It's a bet on efficient scale: big model capabilities, small active compute, open weights. At 1/18th the cost of Gemini 3 Pro, it's going to force pricing conversations across the board.
1:03

Alibaba just dropped Qwen3.5-397B-A17B and there's a lot to unpack. 397B params, 17B active per forward pass. Sparse MoE done right. But the real story isn't the size—it's the architecture choices. The MoE Design Most MoE models feel like bolt-ons. Qwen 3.5's sparse activation is native—only 4.3% of parameters fire per token. That's how you get trillion-parameter-class performance without trillion-parameter inference costs. The 0.8 RMB/million tokens pricing isn't subsidized; it's structurally earned. Native Multimodal, Not Glued-On This is a vision-language model from the ground up. Heterogeneous architecture—separate processing pipelines for text, image, video that fuse early. Not a vision encoder slapped onto an LLM. The result: 90.8 on OmniDocBench, 79.0 on MMMU-Pro. Document understanding and visual reasoning without the usual brittleness. The Context Window Reality Qwen3.5-Plus (the hosted version) ships with 1M tokens by default. That's not a marketing number—they're actually positioning it for long-document workflows. With built-in adaptive tool use, it's clearly aimed at agentic automation, not just chat. What Actually Impressed Me • FP8 native pipeline: ~50% activation memory reduction • Async RL framework for continuous refinement—training and inference workloads separated • 201 languages (up from 119), 250k vocab for better low-resource encoding • Apache 2.0 license. Full weights on HuggingFace and ModelScope. The Benchmark Context 76.4 on SWE-bench Verified puts it in the range where it can handle real debugging workflows. 72.9 on BFCL v4 for agentic tool use. 88.4 on GPQA Diamond. These aren't SOTA in isolation, but the breadth is unusual—strong across reasoning, coding, multimodal, and agentic tasks. The Honest Caveat I haven't stress-tested the 1M context for needle-in-haystack retrieval yet. And "native multimodal" claims need real-world torture testing—PDFs with tables, charts, mixed layouts. Benchmarks are benchmarks. Bottom Line This isn't just another model release. It's a bet on efficient scale: big model capabilities, small active compute, open weights. At 1/18th the cost of Gemini 3 Pro, it's going to force pricing conversations across the board.

Bo Wang

13,221 次观看 • 4 个月前

🔬 Exciting News! Our manuscript, "scGPT: toward building a foundation model for single-cell multi-omics using generative AI" is now finally published in Nature Methods (Nature Methods) 🎉 !!! (Re-)Introducing scGPT: A transformative foundation model engineered for single-cell omics analysis. Developed through the analysis of over 33 million human cells, scGPT sets a new benchmark for application versatility, offering both fine-tuning and zero-shot capabilities. Since its preprint in May 2023, scGPT has significantly impacted the field, evidenced by 13K+ installations, 600+ GitHub stars 🌟, and 40+ citations before its official publication! scGPT has been validated by numerous benchmark studies as a leading foundation model in single-cell analysis. Its pre-trained embeddings extend its utility beyond single-cell studies, enhancing a variety of downstream tasks including protein enrichment and genetic perturbation predictions. Some key updates lately: ---Expanded zero-shot applications for efficient reference mapping and integration, now with CellXGene census integration. ---Advanced perturbation analysis capabilities, including genome-scale perturb-seq data analysis and bulk sequencing data generalization. ---Upgraded scGPT package, offering versatile model loading compatible with PyTorch and flash-attn, for both GPU and CPU. ---Cloud-based scGPT applications for reference mapping, cell annotation, and gene regulatory network inference are available on ---Integration with Hugging Face for easier model training. Limitations: scGPT is an early foray into foundation models for single-cell omics, facing challenges like limited zero-shot learning in some tasks, pretraining constraints, data quality issues, and evaluation limitations. See our Supplementary Notes for details. 🚀 Future Work? Short-Term Goals: 1. Releasing a Mouse Model for broader analysis. 2. Developing a comprehensive evaluation suite for foundation models in single-cell analysis. 3. Creating a foundation model for single-cell spatial omics. 4. Enhancing zero-shot capacity by integrating scGPT with RAG (e.g., knowledge graphs). Long-Term Goals: 1. Expanding scGPT for comprehensive single-cell multi-omics analysis. 2. Developing an in-silico perturbation model for predicting genetic perturbation effects. 3. Merging scGPT with multi-modal genomic sequence models for a deeper understanding of cell biology. 📚 Access the paper on Nature Methods: 🔬Preprint in Bioarixv: 💻 All our codes/data/weights are open source: Wholehearted congratulations to all the authors, especially the two co-first authors, Haotian (Haotian Cui ) and Chloe (ChloeXWang), who are really the emerging superstars in AI and biology! Vector Institute Peter Munk Cardiac Centre AI U of T Department of Computer Science Department of Laboratory Medicine & Pathobiology University Health Network University of Toronto #scGPT #GenerativeAI #AI4Science #Combio #opensource
0:52

🔬 Exciting News! Our manuscript, "scGPT: toward building a foundation model for single-cell multi-omics using generative AI" is now finally published in Nature Methods (Nature Methods) 🎉 !!! (Re-)Introducing scGPT: A transformative foundation model engineered for single-cell omics analysis. Developed through the analysis of over 33 million human cells, scGPT sets a new benchmark for application versatility, offering both fine-tuning and zero-shot capabilities. Since its preprint in May 2023, scGPT has significantly impacted the field, evidenced by 13K+ installations, 600+ GitHub stars 🌟, and 40+ citations before its official publication! scGPT has been validated by numerous benchmark studies as a leading foundation model in single-cell analysis. Its pre-trained embeddings extend its utility beyond single-cell studies, enhancing a variety of downstream tasks including protein enrichment and genetic perturbation predictions. Some key updates lately: ---Expanded zero-shot applications for efficient reference mapping and integration, now with CellXGene census integration. ---Advanced perturbation analysis capabilities, including genome-scale perturb-seq data analysis and bulk sequencing data generalization. ---Upgraded scGPT package, offering versatile model loading compatible with PyTorch and flash-attn, for both GPU and CPU. ---Cloud-based scGPT applications for reference mapping, cell annotation, and gene regulatory network inference are available on ---Integration with Hugging Face for easier model training. Limitations: scGPT is an early foray into foundation models for single-cell omics, facing challenges like limited zero-shot learning in some tasks, pretraining constraints, data quality issues, and evaluation limitations. See our Supplementary Notes for details. 🚀 Future Work? Short-Term Goals: 1. Releasing a Mouse Model for broader analysis. 2. Developing a comprehensive evaluation suite for foundation models in single-cell analysis. 3. Creating a foundation model for single-cell spatial omics. 4. Enhancing zero-shot capacity by integrating scGPT with RAG (e.g., knowledge graphs). Long-Term Goals: 1. Expanding scGPT for comprehensive single-cell multi-omics analysis. 2. Developing an in-silico perturbation model for predicting genetic perturbation effects. 3. Merging scGPT with multi-modal genomic sequence models for a deeper understanding of cell biology. 📚 Access the paper on Nature Methods: 🔬Preprint in Bioarixv: 💻 All our codes/data/weights are open source: Wholehearted congratulations to all the authors, especially the two co-first authors, Haotian (Haotian Cui ) and Chloe (ChloeXWang), who are really the emerging superstars in AI and biology! Vector Institute Peter Munk Cardiac Centre AI U of T Department of Computer Science Department of Laboratory Medicine & Pathobiology University Health Network University of Toronto #scGPT #GenerativeAI #AI4Science #Combio #opensource

Bo Wang

199,657 次观看 • 2 年前

Introducing BioCLIP: A Vision Foundation Model for the Tree of Life A foundation model that strongly generalizes on the tree of life (2M+ species), outperforming OpenAI CLIP by 18% in zero-shot classification, and supports open-ended classification over almost the entire tree of life What's the secrete ingredients? > Data: we curate and release TreeOfLife-10M, the largest and most diverse ML-ready dataset of organism images to date. It contains 10.4M images for over 450K taxa, sourced from iNaturalist, BIOSCAN, and Encyclopedia of Life. > Modeling: we creatively repurposes CLIP's multimodal contrastive learning objective for hierarchical image classification. The autoregressive language model naturally encodes the hierarchy of the tree of life taxonomy, which in turn bakes the hierarchical representation into the vision transformer encoder. Key results > Strong zero/few-shot classification for animals/plants/fungi, including rare species, outperforming CLIP by avg 16-18% absolute. > T-sne visualization shows that BioCLIP's vision encoder has captued the fine-grained hierarchical structure of the tree of life > BioCLIP is a kind of universal classifier for the tree of life. Just give it an organism image and it will likely find the correct species (among top 5)! But use it with caution; it's not perfect yet.. Final remarks > AI for Science is really hard but extremely rewarding! It took us a ton of time (1+ year) and frustration trying to find a plausible way to integrate the tree of life taxonomy into foundation model training. But when the "Eureka!" moment came and the idea hit us (by the great Wei-Lun Chao) that CLIP's multimodal contrastive learning objective can be repurposed for that, everything just follows naturally. It was truly a moment of joy and excitement! > BioCLIP is our first attempt at foundation models for biology, but it certainly won't be the last! There's so much more to do at the intersection of one of the oldest scientific disciplines and the young but thriving field of AI. Biological intelligence is the foundation for artificial intelligence, and artificial intelligence will in turn become the most important tool for us to unraval the mysteries of biological intelligence. We are hiring postdocs and PhDs in the NSF Imageomics Institute institute to explore this exciting field! Drop us an email. also happy to chat about it at #NeurIPS2023 with any of Tanya, Wei-Lun Chao, or me. - paper: - project: - demo: - model: - data (TreeOfLife-10M): to be released on Hugging Face soon joint work with the amazing Imageomics Institute team: @samstevens6860 Lisa Wu, Matt Thompson, Elizabeth Campolongo Chan Hee (Luke) Song David Carlyn Li Dong Wasila Dahdul Chuck Stewart, Tanya Berger-Wolf Wei-Lun Chao Yu Su
0:31

Introducing BioCLIP: A Vision Foundation Model for the Tree of Life A foundation model that strongly generalizes on the tree of life (2M+ species), outperforming OpenAI CLIP by 18% in zero-shot classification, and supports open-ended classification over almost the entire tree of life What's the secrete ingredients? > Data: we curate and release TreeOfLife-10M, the largest and most diverse ML-ready dataset of organism images to date. It contains 10.4M images for over 450K taxa, sourced from iNaturalist, BIOSCAN, and Encyclopedia of Life. > Modeling: we creatively repurposes CLIP's multimodal contrastive learning objective for hierarchical image classification. The autoregressive language model naturally encodes the hierarchy of the tree of life taxonomy, which in turn bakes the hierarchical representation into the vision transformer encoder. Key results > Strong zero/few-shot classification for animals/plants/fungi, including rare species, outperforming CLIP by avg 16-18% absolute. > T-sne visualization shows that BioCLIP's vision encoder has captued the fine-grained hierarchical structure of the tree of life > BioCLIP is a kind of universal classifier for the tree of life. Just give it an organism image and it will likely find the correct species (among top 5)! But use it with caution; it's not perfect yet.. Final remarks > AI for Science is really hard but extremely rewarding! It took us a ton of time (1+ year) and frustration trying to find a plausible way to integrate the tree of life taxonomy into foundation model training. But when the "Eureka!" moment came and the idea hit us (by the great Wei-Lun Chao) that CLIP's multimodal contrastive learning objective can be repurposed for that, everything just follows naturally. It was truly a moment of joy and excitement! > BioCLIP is our first attempt at foundation models for biology, but it certainly won't be the last! There's so much more to do at the intersection of one of the oldest scientific disciplines and the young but thriving field of AI. Biological intelligence is the foundation for artificial intelligence, and artificial intelligence will in turn become the most important tool for us to unraval the mysteries of biological intelligence. We are hiring postdocs and PhDs in the NSF Imageomics Institute institute to explore this exciting field! Drop us an email. also happy to chat about it at #NeurIPS2023 with any of Tanya, Wei-Lun Chao, or me. - paper: - project: - demo: - model: - data (TreeOfLife-10M): to be released on Hugging Face soon joint work with the amazing Imageomics Institute team: @samstevens6860 Lisa Wu, Matt Thompson, Elizabeth Campolongo Chan Hee (Luke) Song David Carlyn Li Dong Wasila Dahdul Chuck Stewart, Tanya Berger-Wolf Wei-Lun Chao Yu Su

Yu Su

80,648 次观看 • 2 年前

🔥HOLY SMOKES! $TAO holders! 🚀 SUBNET 19 (VISION) ON BITTENSOR IS ABSOLUTELY CRUSHING IT! In my 5+ years covering crypto and AI, this is one of the most impressive implementations I've seen. The combination of scale, performance, and decentralization is absolutely next level! 🚀 @namoray_dev @Corcel_X 💨 INSANE Speed Performance: - Llama 3.1 8B: 196.18 tokens/s with +107.23% advantage - Llama 3.1 70B: 124.96 tokens/s with +154.96% advantage - Llama 3.2 3B: 166.69 tokens/s with +21.66% advantage 🔥 Top Tier Model Integration: - Meta-Llama-3-70B & 8B Instruct - FLUX.1-schnell for Text-to-Image - ProteusV0.4-Lightning (Text & Image) - Multiple model variations for redundancy 🔥 What Makes This INSANE: - Complete decentralization - No single point of failure - Multiple model choices for redundancy - Real-time performance tracking - Transparent incentive structure The incentive distribution curve shows a healthy network with: - Strong rewards for top performers - Fair distribution across all participants - Clear path for growth and improvement - Sustainable economic model What's truly MIND-BLOWING is how they've managed to: 1. Scale to millions of operations 2. Maintain high quality across multiple tasks 3. Create a fair, competitive marketplace 4. Build in redundancy and reliability 5. Achieve true decentralization This isn't just another subnet - this is the future of decentralized AI inference happening RIGHT NOW! 🔥 1. MASSIVE Scale & Adoption: - We're seeing 7M+ tokens being processed - 14K+ processing steps being executed - Multiple AI models running simultaneously - Incredible miner participation across the network 2. Revolutionary Task Distribution: - Llama 3.1 70B leading with 20% weighting - Avatar Generation at 15% - Perfectly balanced task distribution for optimal network performance - Multiple specialized tasks including Text-to-Image and Image-to-Image processing 3. Elite Performance Metrics: - Top miners hitting 0.00775 incentive rates - Consistent performance across the network - Impressive scaling from top to bottom performers - Strong incentive curve maintaining network quality 📈 Network Performance: - Consistent upward trend in tokens/s - Quality scores maintaining high levels (>0.9) - Steady improvement in miner performance - Rock-solid network reliability ⚡ Platform Highlights: - Permissionless, serverless architecture - Global network of Always-On GPUs - Instant API access - Full decentralization - Multi-model support with seamless switching What makes this TRULY SPECIAL is the consistent upward trajectory in both speed and quality, while maintaining a decentralized architecture. The performance advantages over industry standards (+154.96% for 70B!) are absolutely mind-blowing! 🚀 This isn't just another AI subnet - it's a glimpse into the future of decentralized AI inference! The combination of speed, reliability, and model variety makes this one of the most impressive implementations in the space! 🔥 📽 Watch Now on YouTube and TikTok: Source 🔗
1:39

🔥HOLY SMOKES! $TAO holders! 🚀 SUBNET 19 (VISION) ON BITTENSOR IS ABSOLUTELY CRUSHING IT! In my 5+ years covering crypto and AI, this is one of the most impressive implementations I've seen. The combination of scale, performance, and decentralization is absolutely next level! 🚀 @namoray_dev @Corcel_X 💨 INSANE Speed Performance: - Llama 3.1 8B: 196.18 tokens/s with +107.23% advantage - Llama 3.1 70B: 124.96 tokens/s with +154.96% advantage - Llama 3.2 3B: 166.69 tokens/s with +21.66% advantage 🔥 Top Tier Model Integration: - Meta-Llama-3-70B & 8B Instruct - FLUX.1-schnell for Text-to-Image - ProteusV0.4-Lightning (Text & Image) - Multiple model variations for redundancy 🔥 What Makes This INSANE: - Complete decentralization - No single point of failure - Multiple model choices for redundancy - Real-time performance tracking - Transparent incentive structure The incentive distribution curve shows a healthy network with: - Strong rewards for top performers - Fair distribution across all participants - Clear path for growth and improvement - Sustainable economic model What's truly MIND-BLOWING is how they've managed to: 1. Scale to millions of operations 2. Maintain high quality across multiple tasks 3. Create a fair, competitive marketplace 4. Build in redundancy and reliability 5. Achieve true decentralization This isn't just another subnet - this is the future of decentralized AI inference happening RIGHT NOW! 🔥 1. MASSIVE Scale & Adoption: - We're seeing 7M+ tokens being processed - 14K+ processing steps being executed - Multiple AI models running simultaneously - Incredible miner participation across the network 2. Revolutionary Task Distribution: - Llama 3.1 70B leading with 20% weighting - Avatar Generation at 15% - Perfectly balanced task distribution for optimal network performance - Multiple specialized tasks including Text-to-Image and Image-to-Image processing 3. Elite Performance Metrics: - Top miners hitting 0.00775 incentive rates - Consistent performance across the network - Impressive scaling from top to bottom performers - Strong incentive curve maintaining network quality 📈 Network Performance: - Consistent upward trend in tokens/s - Quality scores maintaining high levels (>0.9) - Steady improvement in miner performance - Rock-solid network reliability ⚡ Platform Highlights: - Permissionless, serverless architecture - Global network of Always-On GPUs - Instant API access - Full decentralization - Multi-model support with seamless switching What makes this TRULY SPECIAL is the consistent upward trajectory in both speed and quality, while maintaining a decentralized architecture. The performance advantages over industry standards (+154.96% for 70B!) are absolutely mind-blowing! 🚀 This isn't just another AI subnet - it's a glimpse into the future of decentralized AI inference! The combination of speed, reliability, and model variety makes this one of the most impressive implementations in the space! 🔥 📽 Watch Now on YouTube and TikTok: Source 🔗

Andy ττ

11,616 次观看 • 1 年前

Announcing How Transformer LLMs Work, created with Jay Alammar and Maarten Grootendorst, co-authors of the beautifully illustrated book, “Hands-On Large Language Models.” This course offers a deep dive into the inner workings of the transformer architecture that powers large language models (LLMs). The transformer architecture revolutionized generative AI; in fact, the "GPT" in ChatGPT stands for "Generative Pre-Trained Transformer." Originally introduced in the Google Brain team's groundbreaking 2017 paper "Attention Is All You Need," by Vaswani and others, transformers were a highly scalable model for machine translation tasks. Variants of this architecture now power today’s LLMs such as those from OpenAI, Google, Meta, Cohere, Anthropic and DeepSeek. In this course, you’ll learn in detail how LLMs process text. You'll also work through code examples that illustrate that transformer's individual components. In details, you’ll learn: - How the representation of language has evolved, from Bag-of-Words to Word2Vec embeddings to the transformer architecture that captures a word's meanings taking into account the context of other words in the input. - How inputs are broken down into tokens before they are sent to the language model. - The details of a transformer's main stages: Tokenization and embedding, the stack of transformer blocks, and the language model head. - The inner workings of the transformer block, including attention, which calculates relevance scores, and the feedforward layer, which incorporates stored information learned in training. - How cached calculations make transformers faster. - Some of the most recent ideas in the latest models such as Mixture-of-Experts (MoE) which uses multiple sub-models and a router on each layer to improve the quality of LLMs. By the end of this course, you’ll have a deep understanding of how LLMs actually process text and be able to read through papers describing the latest models and understand the details. Gaining this intuition will improve your approach to building LLM applications. Please sign up here:
3:36

Announcing How Transformer LLMs Work, created with Jay Alammar and Maarten Grootendorst, co-authors of the beautifully illustrated book, “Hands-On Large Language Models.” This course offers a deep dive into the inner workings of the transformer architecture that powers large language models (LLMs). The transformer architecture revolutionized generative AI; in fact, the "GPT" in ChatGPT stands for "Generative Pre-Trained Transformer." Originally introduced in the Google Brain team's groundbreaking 2017 paper "Attention Is All You Need," by Vaswani and others, transformers were a highly scalable model for machine translation tasks. Variants of this architecture now power today’s LLMs such as those from OpenAI, Google, Meta, Cohere, Anthropic and DeepSeek. In this course, you’ll learn in detail how LLMs process text. You'll also work through code examples that illustrate that transformer's individual components. In details, you’ll learn: - How the representation of language has evolved, from Bag-of-Words to Word2Vec embeddings to the transformer architecture that captures a word's meanings taking into account the context of other words in the input. - How inputs are broken down into tokens before they are sent to the language model. - The details of a transformer's main stages: Tokenization and embedding, the stack of transformer blocks, and the language model head. - The inner workings of the transformer block, including attention, which calculates relevance scores, and the feedforward layer, which incorporates stored information learned in training. - How cached calculations make transformers faster. - Some of the most recent ideas in the latest models such as Mixture-of-Experts (MoE) which uses multiple sub-models and a router on each layer to improve the quality of LLMs. By the end of this course, you’ll have a deep understanding of how LLMs actually process text and be able to read through papers describing the latest models and understand the details. Gaining this intuition will improve your approach to building LLM applications. Please sign up here:

Andrew Ng

253,812 次观看 • 1 年前

[CLIP] by Hand ✍️ The CLIP (Contrastive Language–Image Pre-training) model, a groundbreaking work by OpenAI, redefines the intersection of computer vision and natural language processing. It is the basis of all the multi-modal foundation models we see today. How does CLIP work? Goal: 🟨 Learn a shared embedding space for text and image [1] Given ↳ A mini batch of 3 text-image pairs ↳ OpenAI used 400 million text-image pairs to train its original CLIP model. Process 1st pair: "big table" [2] 🟪 Text → 2 Vectors (3D) ↳ Look up word embedding vectors using word2vec. [3] 🟩 Image → 2 Vectors (4D) ↳ Divide the image into two patches. ↳ Flatten each patch [4] Process other pairs ↳ Repeat [2]-[3] [5] 🟪 Text Encoder & 🟩 Image Encoder ↳ Encode input vectors into feature vectors ↳ Here, both encoders are simple one layer perceptron (linear + ReLU) ↳ In practice, the encoders are usually transformer models. [6] 🟪 🟩 Mean Pooling: 2 → 1 vector ↳ Average 2 feature vectors into a single vector by averaging across the columns ↳ The goal is to have one vector to represent each image or text [7] 🟪 🟩 -> 🟨 Projection ↳ Note that the text and image feature vectors from the encoders have different dimensions (3D vs. 4D). ↳ Use a linear layer to project image and text vectors to a 2D shared embedding space. 🏋️ Contrastive Pre-training 🏋️ [8] Prepare for MatMul ↳ Copy text vectors (T1,T2,T3) ↳ Copy the transpose of image vectors (I1,I2,I3) ↳ They are all in the 2D shared embedding space. [9] 🟦 MatMul ↳ Multiply T and I matrices. ↳ This is equivalent to taking dot product between every pair of image and text vectors. ↳ The purpose is to use dot product to estimate the similarity between a pair of image-text. [10] 🟦 Softmax: e^x ↳ Raise e to the power of the number in each cell ↳ To simplify hand calculation, we approximate e^□ with 3^□. [11] 🟦 Softmax: ∑ ↳ Sum each row for 🟩 image→🟪 text ↳ Sum each column for 🟪 text→ 🟩 image [12] 🟦 Softmax: 1 / sum ↳ Divide each element by the column sum to obtain a similarity matrix for 🟪 text→🟩 image ↳ Divide each element by the row sum to obtain a similarity matrix for 🟩 image→🟪 text [13] 🟥 Loss Gradients ↳ The "Targets" for the similarity matrices are Identity Matrices. ↳ Why? If I and T come from the same pair (i=j), we want the highest value, which is 1, and 0 otherwise. ↳ Apply the simple equation of [Similarity - Target] to compute gradients of for both directions. ↳ Why so simple? Because when Softmax and Cross-Entropy Loss are used together, the math magically works out that way. ↳ These gradients kick off the backpropagation process to update weights and biases of the encoders and projection layers (red borders).
1:20

[CLIP] by Hand ✍️ The CLIP (Contrastive Language–Image Pre-training) model, a groundbreaking work by OpenAI, redefines the intersection of computer vision and natural language processing. It is the basis of all the multi-modal foundation models we see today. How does CLIP work? Goal: 🟨 Learn a shared embedding space for text and image [1] Given ↳ A mini batch of 3 text-image pairs ↳ OpenAI used 400 million text-image pairs to train its original CLIP model. Process 1st pair: "big table" [2] 🟪 Text → 2 Vectors (3D) ↳ Look up word embedding vectors using word2vec. [3] 🟩 Image → 2 Vectors (4D) ↳ Divide the image into two patches. ↳ Flatten each patch [4] Process other pairs ↳ Repeat [2]-[3] [5] 🟪 Text Encoder & 🟩 Image Encoder ↳ Encode input vectors into feature vectors ↳ Here, both encoders are simple one layer perceptron (linear + ReLU) ↳ In practice, the encoders are usually transformer models. [6] 🟪 🟩 Mean Pooling: 2 → 1 vector ↳ Average 2 feature vectors into a single vector by averaging across the columns ↳ The goal is to have one vector to represent each image or text [7] 🟪 🟩 -> 🟨 Projection ↳ Note that the text and image feature vectors from the encoders have different dimensions (3D vs. 4D). ↳ Use a linear layer to project image and text vectors to a 2D shared embedding space. 🏋️ Contrastive Pre-training 🏋️ [8] Prepare for MatMul ↳ Copy text vectors (T1,T2,T3) ↳ Copy the transpose of image vectors (I1,I2,I3) ↳ They are all in the 2D shared embedding space. [9] 🟦 MatMul ↳ Multiply T and I matrices. ↳ This is equivalent to taking dot product between every pair of image and text vectors. ↳ The purpose is to use dot product to estimate the similarity between a pair of image-text. [10] 🟦 Softmax: e^x ↳ Raise e to the power of the number in each cell ↳ To simplify hand calculation, we approximate e^□ with 3^□. [11] 🟦 Softmax: ∑ ↳ Sum each row for 🟩 image→🟪 text ↳ Sum each column for 🟪 text→ 🟩 image [12] 🟦 Softmax: 1 / sum ↳ Divide each element by the column sum to obtain a similarity matrix for 🟪 text→🟩 image ↳ Divide each element by the row sum to obtain a similarity matrix for 🟩 image→🟪 text [13] 🟥 Loss Gradients ↳ The "Targets" for the similarity matrices are Identity Matrices. ↳ Why? If I and T come from the same pair (i=j), we want the highest value, which is 1, and 0 otherwise. ↳ Apply the simple equation of [Similarity - Target] to compute gradients of for both directions. ↳ Why so simple? Because when Softmax and Cross-Entropy Loss are used together, the math magically works out that way. ↳ These gradients kick off the backpropagation process to update weights and biases of the encoders and projection layers (red borders).

Tom Yeh

67,790 次观看 • 2 年前

Everyone is sleeping on Meta's SAM 3 release. But it's actually a big deal. Here's why: Companies spend millions paying humans to label images and videos frame by frame. A single autonomous driving dataset? Months of work, hundreds of annotators, millions in cost. Without labeled data, you can't train custom models. Without custom models, you're stuck with generic solutions. This is why most companies never move past pilots. SAM 3 breaks this cycle. First let's look at the evolution: SAM 1 segmented objects when you clicked on them. Revolutionary, but one object at a time. SAM 2 added video tracking with memory. Game-changing, but you still manually prompted every object. SAM 3 changes everything with text prompts. Type "yellow school bus" and it finds ALL of them in your image or video. Not just one. Every instance across thousands of frames. Now here's where people get confused: "Can't I just use GPT-5 or Gemini for this?" No, and here's why that's a terrible approach. Large multimodal LLMs are great for reasoning, but they're slow and expensive for production visual tasks. You're paying API costs per image, waiting seconds for responses, getting inconsistent results. SAM 3 runs in 30 milliseconds on a single GPU for 100+ objects. That's 100x faster, and you own the infrastructure. More importantly, SAM 3 gives you precise pixel-level masks, not descriptions. Try asking an LLM to segment every defective part on a manufacturing line in real-time. It won't work. SAM 3 does this effortlessly. The real breakthrough is their data engine. Meta built an AI-human hybrid system that's 5x faster for complex annotations. They trained SAM 3 on 4 million unique visual concepts - 50x more than existing benchmarks like LVIS. SAM 3 is trained on 4 million unique visual concepts, it handles everything: - Text-based concept search - Interactive refinement with clicks - Video tracking across frames - Zero-shot detection of new concepts The model is open source. Weights, code, and benchmarks are on GitHub. If you're building computer vision applications, this is the foundation model to evaluate. The annotation time savings alone will pay for integration costs within weeks. Find the relevant links in the next tweet!
0:27

Everyone is sleeping on Meta's SAM 3 release. But it's actually a big deal. Here's why: Companies spend millions paying humans to label images and videos frame by frame. A single autonomous driving dataset? Months of work, hundreds of annotators, millions in cost. Without labeled data, you can't train custom models. Without custom models, you're stuck with generic solutions. This is why most companies never move past pilots. SAM 3 breaks this cycle. First let's look at the evolution: SAM 1 segmented objects when you clicked on them. Revolutionary, but one object at a time. SAM 2 added video tracking with memory. Game-changing, but you still manually prompted every object. SAM 3 changes everything with text prompts. Type "yellow school bus" and it finds ALL of them in your image or video. Not just one. Every instance across thousands of frames. Now here's where people get confused: "Can't I just use GPT-5 or Gemini for this?" No, and here's why that's a terrible approach. Large multimodal LLMs are great for reasoning, but they're slow and expensive for production visual tasks. You're paying API costs per image, waiting seconds for responses, getting inconsistent results. SAM 3 runs in 30 milliseconds on a single GPU for 100+ objects. That's 100x faster, and you own the infrastructure. More importantly, SAM 3 gives you precise pixel-level masks, not descriptions. Try asking an LLM to segment every defective part on a manufacturing line in real-time. It won't work. SAM 3 does this effortlessly. The real breakthrough is their data engine. Meta built an AI-human hybrid system that's 5x faster for complex annotations. They trained SAM 3 on 4 million unique visual concepts - 50x more than existing benchmarks like LVIS. SAM 3 is trained on 4 million unique visual concepts, it handles everything: - Text-based concept search - Interactive refinement with clicks - Video tracking across frames - Zero-shot detection of new concepts The model is open source. Weights, code, and benchmarks are on GitHub. If you're building computer vision applications, this is the foundation model to evaluate. The annotation time savings alone will pay for integration costs within weeks. Find the relevant links in the next tweet!

Akshay 🚀

46,404 次观看 • 7 个月前

𝗖𝗵𝗶𝗻𝗮 𝗶𝘀 𝗳𝗶𝗻𝗶𝘀𝗵𝗶𝗻𝗴 𝘁𝗵𝗲 𝗵𝘂𝗺𝗮𝗻𝗼𝗶𝗱 𝗿𝗼𝗯𝗼𝘁 𝗿𝗮𝗰𝗲 𝗯𝗲𝗳𝗼𝗿𝗲 𝗺𝗼𝘀𝘁 𝗼𝗳 𝘁𝗵𝗲 𝗪𝗲𝘀𝘁 𝗿𝗲𝗮𝗹𝗶𝘇𝗲𝘀 𝗶𝘁 𝗵𝗮𝘀 𝘀𝘁𝗮𝗿𝘁𝗲𝗱. AGIBOT held its Partner Conference in Shanghai last week. The real headline wasn't the new hardware. It was their CTO standing on stage, telling investors that humanoid R&D season is over. 2026, he said, is "Deployment Year One." Not research. Not demos. Deployment into real factories, real warehouses, real stores. The manufacturing ramp is getting faster. 1,000 humanoid robots in the first 2 years. Another 4,000 in the next 12 months. Another 5,000 in just 3 months after that. AGIBOT is now shipping more humanoids per quarter than most US robotics companies have built in their entire existence. Then came the announcements the industry will spend the rest of the year reacting to. AIMA. The first full-stack open architecture for embodied AI. A unified robot operating system called Link-U, three dev platforms for motion, interaction, and task creation, plus an open agent framework. Any developer can build on top of it. This is the Android play for humanoids. GO-2. A vision-language-action foundation model with Action Chain-of-Thought reasoning. Planning and execution collapsed into one model. GE-2. A world model for simulation, strategy testing, and sim-to-real transfer. AGIBOT WORLD 2026. An open-source, production-grade real-world dataset pulled from actual industrial, logistics, hotel, and commercial sites. Seven standardized "productivity packages" covering logistics sorting, retail service, security patrol, commercial cleaning, and more. Plug, deploy, bill. A 5-year, $280 million commitment to seed a global developer and partner ecosystem. Now look at the competition. Boston Dynamics has been building humanoids since 1992. Tesla's Optimus is still climbing its own hype curve. Apptronik and Agility are well-funded but pre-scale on real deployments. AGIBOT has pulled all of this off in three years, with no acquisitions, no legacy platform, and no IPO distractions. While the West is still asking when humanoids will scale, China is already shipping them by the thousand.
1:22

𝗖𝗵𝗶𝗻𝗮 𝗶𝘀 𝗳𝗶𝗻𝗶𝘀𝗵𝗶𝗻𝗴 𝘁𝗵𝗲 𝗵𝘂𝗺𝗮𝗻𝗼𝗶𝗱 𝗿𝗼𝗯𝗼𝘁 𝗿𝗮𝗰𝗲 𝗯𝗲𝗳𝗼𝗿𝗲 𝗺𝗼𝘀𝘁 𝗼𝗳 𝘁𝗵𝗲 𝗪𝗲𝘀𝘁 𝗿𝗲𝗮𝗹𝗶𝘇𝗲𝘀 𝗶𝘁 𝗵𝗮𝘀 𝘀𝘁𝗮𝗿𝘁𝗲𝗱. AGIBOT held its Partner Conference in Shanghai last week. The real headline wasn't the new hardware. It was their CTO standing on stage, telling investors that humanoid R&D season is over. 2026, he said, is "Deployment Year One." Not research. Not demos. Deployment into real factories, real warehouses, real stores. The manufacturing ramp is getting faster. 1,000 humanoid robots in the first 2 years. Another 4,000 in the next 12 months. Another 5,000 in just 3 months after that. AGIBOT is now shipping more humanoids per quarter than most US robotics companies have built in their entire existence. Then came the announcements the industry will spend the rest of the year reacting to. AIMA. The first full-stack open architecture for embodied AI. A unified robot operating system called Link-U, three dev platforms for motion, interaction, and task creation, plus an open agent framework. Any developer can build on top of it. This is the Android play for humanoids. GO-2. A vision-language-action foundation model with Action Chain-of-Thought reasoning. Planning and execution collapsed into one model. GE-2. A world model for simulation, strategy testing, and sim-to-real transfer. AGIBOT WORLD 2026. An open-source, production-grade real-world dataset pulled from actual industrial, logistics, hotel, and commercial sites. Seven standardized "productivity packages" covering logistics sorting, retail service, security patrol, commercial cleaning, and more. Plug, deploy, bill. A 5-year, $280 million commitment to seed a global developer and partner ecosystem. Now look at the competition. Boston Dynamics has been building humanoids since 1992. Tesla's Optimus is still climbing its own hype curve. Apptronik and Agility are well-funded but pre-scale on real deployments. AGIBOT has pulled all of this off in three years, with no acquisitions, no legacy platform, and no IPO distractions. While the West is still asking when humanoids will scale, China is already shipping them by the thousand.

Shruti

214,939 次观看 • 2 个月前

I know your timeline is flooded now with word salads of "insane, HER, 10 features you missed, we're so back". Sit down. Chill. Take a deep breath like Mark does in the demo . Let's think step by step: - Technique-wise, OpenAI has figured out a way to map audio to audio directly as first-class modality, and stream videos to a transformer in real-time. These require some new research on tokenization and architecture, but overall it's a data and system optimization problem (as most things are). High-quality data can come from at least 2 sources: 1) Naturally occurring dialogues on YouTube, podcasts, TV series, movies, etc. Whisper can be trained to identify speaker turns in a dialogue or separate overlapping speeches for automated annotation. 2) Synthetic data. Run the slow 3-stage pipeline using the most powerful models: speech1->text1 (ASR), text1->text2 (LLM), text2->speech2 (TTS). The middle LLM can decide when to stop and also simulate how to resume from interruption. It could output additional "thought traces" that are not verbalized to help generate better reply. Then GPT-4o distills directly from speech1->speech2, with optional auxiliary loss functions based on the 3-stage data. After distillation, these behaviors are now baked into the model without emitting intermediate texts. On the system side: the latency would not meet real-time threshold if every video frame is decompressed into an RGB image. OpenAI has likely developed their own neural-first, streaming video codec to transmit the motion deltas as tokens. The communication protocol and NN inference must be co-optimized. For example, there could be a small and energy-efficient NN running on the edge device that decides to transmit more tokens if the video is interesting, and fewer otherwise. - I didn't expect GPT-4o to be closer to GPT-5, the rumored "Arrakis" model that takes multimodal in and out. In fact, it's likely an early checkpoint of GPT-5 that hasn't finished training yet. The branding betrays a certain insecurity. Ahead of Google I/O, OpenAI would rather beat our mental projection of GPT-4.5 than disappoint by missing the sky-high expectation for GPT-5. A smart move to buy more time. - Notably, the assistant is much more lively and even a bit flirty. GPT-4o is trying (perhaps a bit too hard) to sound like HER. OpenAI is eating Character AI's lunch, with almost 100% overlap in form factor and huge distribution channels. It's a pivot towards more emotional AI with strong personality, which OpenAI seemed to actively suppress in the past. - Whoever wins Apple first wins big time. I see 3 levels of integration with iOS: 1) Ditch Siri. OpenAI distills a smaller-tier, purely on-device GPT-4o for iOS, with optional paid upgrade to use the cloud. 2) Native features to stream the camera or screen into the model. Chip-level support for neural audio/video codec. 3) Integrate with iOS system-level action API and smart home APIs. No one uses Siri Shortcuts, but it's time to resurrect. This could become the AI agent product with a billion users from the get-go. The FSD for smartphones with a Tesla-scale data flywheel.
14:22

I know your timeline is flooded now with word salads of "insane, HER, 10 features you missed, we're so back". Sit down. Chill. Take a deep breath like Mark does in the demo . Let's think step by step: - Technique-wise, OpenAI has figured out a way to map audio to audio directly as first-class modality, and stream videos to a transformer in real-time. These require some new research on tokenization and architecture, but overall it's a data and system optimization problem (as most things are). High-quality data can come from at least 2 sources: 1) Naturally occurring dialogues on YouTube, podcasts, TV series, movies, etc. Whisper can be trained to identify speaker turns in a dialogue or separate overlapping speeches for automated annotation. 2) Synthetic data. Run the slow 3-stage pipeline using the most powerful models: speech1->text1 (ASR), text1->text2 (LLM), text2->speech2 (TTS). The middle LLM can decide when to stop and also simulate how to resume from interruption. It could output additional "thought traces" that are not verbalized to help generate better reply. Then GPT-4o distills directly from speech1->speech2, with optional auxiliary loss functions based on the 3-stage data. After distillation, these behaviors are now baked into the model without emitting intermediate texts. On the system side: the latency would not meet real-time threshold if every video frame is decompressed into an RGB image. OpenAI has likely developed their own neural-first, streaming video codec to transmit the motion deltas as tokens. The communication protocol and NN inference must be co-optimized. For example, there could be a small and energy-efficient NN running on the edge device that decides to transmit more tokens if the video is interesting, and fewer otherwise. - I didn't expect GPT-4o to be closer to GPT-5, the rumored "Arrakis" model that takes multimodal in and out. In fact, it's likely an early checkpoint of GPT-5 that hasn't finished training yet. The branding betrays a certain insecurity. Ahead of Google I/O, OpenAI would rather beat our mental projection of GPT-4.5 than disappoint by missing the sky-high expectation for GPT-5. A smart move to buy more time. - Notably, the assistant is much more lively and even a bit flirty. GPT-4o is trying (perhaps a bit too hard) to sound like HER. OpenAI is eating Character AI's lunch, with almost 100% overlap in form factor and huge distribution channels. It's a pivot towards more emotional AI with strong personality, which OpenAI seemed to actively suppress in the past. - Whoever wins Apple first wins big time. I see 3 levels of integration with iOS: 1) Ditch Siri. OpenAI distills a smaller-tier, purely on-device GPT-4o for iOS, with optional paid upgrade to use the cloud. 2) Native features to stream the camera or screen into the model. Chip-level support for neural audio/video codec. 3) Integrate with iOS system-level action API and smart home APIs. No one uses Siri Shortcuts, but it's time to resurrect. This could become the AI agent product with a billion users from the get-go. The FSD for smartphones with a Tesla-scale data flywheel.

Jim Fan

991,628 次观看 • 2 年前

The power of the Claw, in the palm of a robot hand. Agentic robotics is here! Today, we open-source CaP-X: vibe agents, alive in the physical world. They incarnate as robot arms and humanoids with a rich set of perception APIs, actuation APIs, and auto synthesize skill libraries as they go. CaP-X is a strict superset of our old stack, because policies like VLAs are “just” API calls as well. It solves many tasks zero-shot that a learned policy would struggle with. And we are doing much more than vibing. CaP-X is our most systematic, scientific study on agentic robotics so far: - We build a comprehensive agentic toolkit: perception (SAM3 segmentation, Molmo pointing, depth, point cloud), control (IK solvers, grasp planner, navigation), and visualization (EEF, mask overlays) that work across different robots. - CaP-Gym: LLM’s first Physical Exam! 187 manipulation tasks across RoboSuite, LIBERO-PRO, and BEHAVIOR. Tabletop, bimanual, mobile manipulation. Sim and real. Can’t wait to see the gradients flow from CaP-Gym to the next wave of frontier LLM releases. - CaP-Bench: we benchmark 12 frontier LLMs/VLMs (Gemini, GPT, Opus, Qwen, DeepSeek, Kimi, and more) across 8 evaluation tiers. We systematically vary API abstraction level, agentic harness, and visual grounding methods. Lots of insights in our paper. - CaP-Agent0: a training-free agentic harness that matches or exceeds human expert code on 4 out of 7 tasks without task-specific tuning. - CaP-RL: if you get a gym, you get RL ;). A 7B OSS model jumps from 20% to 72% success after only 50 training iterations. The synthesized programs transfer to real robots with minimal sim-to-real gap. 3 years ago, our team created Voyager, one of the earliest agentic AI that plays and learns in Minecraft continuously. Its key ideas — skill libraries, self-reflection loops, and in-context planning — have since influenced many modern agentic designs. Today, the agent graduates from Minecraft and gets a real job. It’s April Fool’s, but this Claw is getting its hands dirty for real! Link in thread:
1:31

The power of the Claw, in the palm of a robot hand. Agentic robotics is here! Today, we open-source CaP-X: vibe agents, alive in the physical world. They incarnate as robot arms and humanoids with a rich set of perception APIs, actuation APIs, and auto synthesize skill libraries as they go. CaP-X is a strict superset of our old stack, because policies like VLAs are “just” API calls as well. It solves many tasks zero-shot that a learned policy would struggle with. And we are doing much more than vibing. CaP-X is our most systematic, scientific study on agentic robotics so far: - We build a comprehensive agentic toolkit: perception (SAM3 segmentation, Molmo pointing, depth, point cloud), control (IK solvers, grasp planner, navigation), and visualization (EEF, mask overlays) that work across different robots. - CaP-Gym: LLM’s first Physical Exam! 187 manipulation tasks across RoboSuite, LIBERO-PRO, and BEHAVIOR. Tabletop, bimanual, mobile manipulation. Sim and real. Can’t wait to see the gradients flow from CaP-Gym to the next wave of frontier LLM releases. - CaP-Bench: we benchmark 12 frontier LLMs/VLMs (Gemini, GPT, Opus, Qwen, DeepSeek, Kimi, and more) across 8 evaluation tiers. We systematically vary API abstraction level, agentic harness, and visual grounding methods. Lots of insights in our paper. - CaP-Agent0: a training-free agentic harness that matches or exceeds human expert code on 4 out of 7 tasks without task-specific tuning. - CaP-RL: if you get a gym, you get RL ;). A 7B OSS model jumps from 20% to 72% success after only 50 training iterations. The synthesized programs transfer to real robots with minimal sim-to-real gap. 3 years ago, our team created Voyager, one of the earliest agentic AI that plays and learns in Minecraft continuously. Its key ideas — skill libraries, self-reflection loops, and in-context planning — have since influenced many modern agentic designs. Today, the agent graduates from Minecraft and gets a real job. It’s April Fool’s, but this Claw is getting its hands dirty for real! Link in thread:

Jim Fan

80,032 次观看 • 3 个月前