Відео

Funny comedy piece, with all those nonsensical analogies, such as "compressing code" :D

really appreciate your efforts

I'm very disappointed it got worse on tool usage... It's literally what we need most for agents and the least consistent thing llama 3.1 did

Ok crazy !! Just in time 😂 i’m building a software and I was looking for something powerful but still cheap if it's not free knowing I’m still in the MVP development! 🙌🏽 fire 🔥

Meta AI just released Llama 3.3, an open-source language model designed to offer better performance and quality for text-based applications, like synthetic data generation, at a much lower cost. Llama 3.3 tackles some of the key challenges in the NLP space by providing a more affordable and easier-to-use solution. The improvements in this version are mainly due to a new alignment process and advances in online reinforcement learning. Essentially, Llama 3.3 delivers performance similar to its predecessor, Llama 3.1-405B, but in a smaller, 70-billion parameter model that can run on regular developer hardware. This makes advanced AI capabilities more accessible to a wider audience. Llama 3.3 comes with several technical upgrades that boost its practicality. One of the major enhancements is the reduction in the number of parameters-from 405 billion in Llama 3.1 to just 70 billion-without sacrificing performance. This was achieved through online preference optimization and better alignment during the training process. The model’s alignment with user preferences, powered by reinforcement learning, means it can generate more relevant and context-aware responses. The smaller size also makes it easier to deploy, as it requires less computational power and memory. Developers can now run Llama 3.3 on their personal computers instead of relying on expensive GPUs or cloud infrastructure, which significantly broadens access to high-quality NLP tools. Read the full article here: www.marktechpost.com/2024/12/06/meta-ai-just-open-sourced-llama-3-3-a-new-70b-multilingual-large-language-model-llm/ Download the Model on Hugging Face: huggingface.co/meta-llama/Llama-3.3-70B-Instruct Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 60k+ ML SubReddit: www.reddit.com/r/machinelearningnews/ ⚓ Feel free to subscribe to our AI Research Newsletter read by 30k+ AI and Data Professionals: airesearchinsights.com/subscribe

Ruliad AI released Deepthought-8B-LLaMA-v0.01-alpha, focusing on reasoning transparency and control. This model, built on LLaMA-3.1 with 8 billion parameters, is designed to offer sophisticated problem-solving capabilities comparable to much larger models while maintaining operational efficiency. Deepthought-8B distinguishes itself with unique features aimed at making AI reasoning more accessible and understandable. The standout characteristic is its transparent reasoning mechanism, where every step in the decision-making process is documented. This feature ensures users can follow the model’s thought process, outputted in a structured JSON format. This step-by-step reasoning builds trust in its outputs and facilitates seamless integration into applications requiring clear and explainable AI logic. Another aspect of Deepthought-8B is its programmable reasoning patterns. Unlike many models that require retraining for different tasks, this model allows customization of reasoning approaches without necessitating retraining. This adaptability makes it suitable for various applications, from coding tasks to complex problem-solving scenarios. Also, its scalability in test-time computing ensures it can adjust reasoning depth based on the complexity of tasks, providing users with a versatile tool for various challenges.... Read the full article here: www.marktechpost.com/2024/12/06/ruliad-ai-releases-deepthought-8b-a-new-small-language-model-built-on-llama-3-1-with-test-time-compute-scaling-and-deliverers-transparent-reasoning/ Model Weights are available on Hugging Face: huggingface.co/ruliad/deepthought-8b-llama-v0.01-alpha Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 60k+ ML SubReddit: www.reddit.com/r/machinelearningnews/ ⚓ Feel free to subscribe to our AI Research Newsletter read by 30k+ AI and Data Professionals: airesearchinsights.com/subscribe

Google DeepMind Just Released PaliGemma 2: A New Family of Open-Weight Vision Language Models (3B, 10B and 28B) recently introduced the PaliGemma 2 series, a new family of Vision-Language Models (VLMs) with parameter sizes of 3 billion (3B), 10 billion (10B), and 28 billion (28B). The models support resolutions of 224×224, 448×448, and 896×896 pixels. This release includes nine pre-trained models with different combinations of sizes and resolutions, making them versatile for a variety of use cases. Two of these models are also fine-tuned on the DOCCI dataset, which contains image-text caption pairs, and support parameter sizes of 3B and 10B at a resolution of 448×448 pixels. Since these models are open-weight, they can be easily adopted as a direct replacement or upgrade for the original PaliGemma, offering users more flexibility for transfer learning and fine-tuning.... Read the full article here: www.marktechpost.com/2024/12/05/google-ai-just-released-paligemma-2-a-new-family-of-open-weight-vision-language-models-3b-10b-and-28b/ Paper: arxiv.org/abs/2412.03555 Models on Hugging Face: huggingface.co/collections/google/paligemma-2-release-67500e1e1dbfdd4dee27ba48 Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 60k+ ML SubReddit: www.reddit.com/r/machinelearningnews/ ⚓ Feel free to subscribe to our AI Research Newsletter read by 30k+ AI and Data Professionals: airesearchinsights.com/subscribe

I am not able to get it working with Azure. I set the keys as expected but I get the 404 Resource not found error. I use the gpt-4o-mini model with version number 2024-05-01-preview. Also tried with 2024-07-18 as the version. I set the following keys as the first line of code import os os.environ["AZURE_OPENAI_API_KEY"] = "<key>" os.environ["AZURE_OPENAI_ENDPOINT"] = "<endpoing url>" os.environ["AZURE_API_VERSION"] = "2024-05-01-preview" Also in the examples\config.ini file, did these.. API_TYPE=azure AZURE_API_VERSION="2024-05-01-preview" In azure, the deployment name is "daa-gpt-4o-mini" for gpt-4o-mini model. It is failing in the customer = factory.generate_person() method. Any help on this would be highly appreciated. Let me know if you need more info..

Thank you so much for this walkthrough!! :) Made my day!

Hawk tuah AI podcast

This is such a great explanation!!

The Allen Institute for AI (AI2) has announced the release of Tülu 3, a state-of-the-art family of instruction-following models designed to set a new benchmark in AI capabilities. This release includes state-of-the-art features, methodologies, and tools, providing researchers and developers with a comprehensive, open-source solution. With Tülu 3, AI2 has successfully addressed a broad range of tasks, from conversational AI to complex problem-solving domains such as mathematics, reasoning, and evaluation. Tülu 3 is a model family prioritizing transparency, openness, and state-of-the-art performance. The models are based on Meta’s Llama 3.1 framework and have been fine-tuned on an extensive dataset mix comprising publicly available, synthetic, and human-created data. This approach ensures that Tülu 3 achieves excellence across diverse tasks, including specialized domains like MATH, GSM8K, and IFEval while maintaining strong capabilities in general-purpose chat and reasoning tasks. Read the full article here: www.marktechpost.com/2024/11/21/the-allen-institute-for-ai-ai2-releases-tulu-3-a-set-of-state-of-the-art-instruct-models-with-fully-open-data-eval-code-and-training-algorithms/ Tülu 3 8B (Llama-3.1-Tulu-3-8B): huggingface.co/allenai/Llama-3.1-Tulu-3-8B Tülu 3 70B (Llama-3.1-Tulu-3-70B): huggingface.co/allenai/Llama-3.1-Tulu-3-70B Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/ ⚓ Feel free to subscribe to our AI Research Newsletter read by 30k+ AI and Data Professionals: airesearchinsights.com/subscribe

It’s 2024 and I’m listening to two AI’s discuss software that simulates a community of AI’s interacting. Amazing and spooky…

Don't do this guys. 😢

I’m disappointed in myself for not realizing this was AI generated until the last 2 minutes.

DeepSeek has introduced DeepSeek-R1-Lite-Preview, an AI model that boasts both improved performance and increased transparency in reasoning. This model provides a step-by-step explanation of its thought process, addressing a major shortcoming of many AI models that fail to provide insights into their reasoning. The model has demonstrated performance comparable to OpenAI's o1 preview on complex reasoning tasks, particularly in mathematics. By offering a transparent approach to reasoning, DeepSeek-R1-Lite-Preview aims to increase user trust and understanding of AI models while also making these advanced technologies more accessible through open-source release. Read the full article here: www.marktechpost.com/2024/11/20/deepseek-introduces-deepseek-r1-lite-preview-with-complete-reasoning-outputs-matching-openai-o1/ Try it here: chat.deepseek.com/ Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/ ⚓ Feel free to subscribe to our AI Research Newsletter read by 30k+ AI and Data Professionals: airesearchinsights.com/subscribe

What a time to be alive!

Clearly, Meta's NotebooLM isn't as good if you have to use Google's to announce it... Love the irony but was looking to hear the quality

A team of MIT researchers has introduced Boltz-1, the first open-source and commercially accessible model that matches AlphaFold3-level accuracy in predicting biomolecular complexes. Unlike its predecessors, Boltz-1 is fully accessible to the public, with the model weights, training, and inference code released under the MIT license. This openness aims to foster global collaboration and advance biomolecular modeling. Boltz-1 follows the general framework used in AlphaFold3 but introduces several architectural and procedural innovations, including new multiple sequence alignment (MSA) pairing algorithms, a unified cropping approach for efficient training, and an enhanced confidence model. These innovations allow Boltz-1 to deliver high accuracy while remaining accessible and significantly lowering the computational burden. Read the full article here: www.marktechpost.com/2024/11/17/mit-researchers-propose-boltz-1-the-first-open-source-ai-model-achieving-alphafold3-level-accuracy-in-biomolecular-structure-prediction/ Paper: gcorso.github.io/assets/boltz1.pdf Model and Codes: github.com/jwohlwend/boltz Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/ ⚓ Feel free to subscribe to our AI Research Newsletter read by 30k+ AI and Data Professionals: airesearchinsights.com/subscribe

Human acts like AI. rediculous.

Learned my about models, taught by models, by models

Nexa AI Releases OmniVision-968M: World’s Smallest Vision Language Model with 9x Tokens Reduction for Edge Devices. OmniVision-968M has been engineered with improved architecture over LLaVA (Large Language and Vision Assistant), achieving a new level of compactness and efficiency, ideal for running on the edge. With a design focused on the reduction of image tokens by a factor of nine-from 729 to just 81-the latency and computational burden typically associated with such models have been drastically minimized. Read the full article here: www.marktechpost.com/2024/11/15/nexa-ai-releases-omnivision-968m-worlds-smallest-vision-language-model-with-9x-tokens-reduction-for-edge-devices/ Model on Hugging Face: huggingface.co/NexaAIDev/omnivision-968M Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/ ⚓ Feel free to subscribe to our AI Research Newsletter read by 30k+ AI and Data Professionals: airesearchinsights.com/subscribe

Microsoft's TinyTroupe is an experimental Python library that uses large language models to simulate human-like agents with distinct personalities, interests, and goals. The library aims to overcome the limitations of traditional rule-based systems by allowing agents to respond contextually, hold conversations, and even make plans. TinyTroupe has potential applications in education, business, and research, enabling the creation of realistic social simulations that can be used to study group dynamics and emergent behaviors. This advancement in multi-agent simulation could lead to more empathetic and relatable machine systems... Read the full article here: www.marktechpost.com/2024/11/14/microsoft-ai-open-sources-tinytroupe-a-new-python-library-for-llm-powered-multiagent-simulation/ GitHub Page: github.com/microsoft/TinyTroupe?tab=readme-ov-file#tinytroupe- Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/ ⚓ Feel free to subscribe to our AI Research Newsletter read by 30k+ AI and Data Professionals: airesearchinsights.com/subscribe

Nexusflow introduces Athene-V2: an open 72-billion-parameter model suite that aims to address this shift in AI development. Athene-V2 is comparable to OpenAI’s GPT-4o across various benchmarks, offering a specialized, cutting-edge approach to solving real-world problems. This suite includes two distinctive models: Athene-V2-Chat and Athene-V2-Agent, each optimized for specific capabilities. The introduction of Athene-V2 aims to break through the current limitations by offering tailored functionality through focused post-training, making LLMs more efficient and usable in practical settings... Read the full article here: www.marktechpost.com/2024/11/14/nexusflow-releases-athene-v2-an-open-72b-model-suite-comparable-to-gpt-4o-across-benchmarks/ Athene-V2-Chat Model on Hugging Face: huggingface.co/Nexusflow/Athene-V2-Chat Athene-V2-Agent Model on Hugging Face: huggingface.co/Nexusflow/Athene-V2-Agent ⚓ Feel free to subscribe to our AI Research Newsletter read by 30k+ AI and Data Professionals: airesearchinsights.com/subscribe

Read the full article: www.marktechpost.com/2024/11/14/microsoft-released-llm2clip-a-new-ai-technique-in-which-a-llm-acts-as-a-teacher-for-clips-visual-encoder/ Paper: arxiv.org/abs/2411.04997 Models on Hugging Face: huggingface.co/collections/microsoft/llm2clip-672323a266173cfa40b32d4c GitHub: github.com/microsoft/LLM2CLIP/ Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/

Well explain thanks for going in to the details of this great paper

Read the full article here: www.marktechpost.com/2024/11/13/fixie-ai-introduces-ultravox-v0-4-1-a-family-of-open-speech-models-trained-specifically-for-enabling-real-time-conversation-with-llms-and-an-open-weight-alternative-to-gpt-4o-realtime/ Details: www.ultravox.ai/blog/ultravox-an-open-weight-alternative-to-gpt-4o-realtime Models on Hugging Face: huggingface.co/fixie-ai GitHub Page: github.com/fixie-ai/ultravox/ Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/

that notebook lm generated podcast!

Read the full article here: www.marktechpost.com/2024/11/11/deepmind-released-alphafold-3-inference-codebase-model-weights-and-an-on-demand-server/ Paper: www.nature.com/articles/s41586-024-07487-w Codebase: github.com/google-deepmind/alphafold3?tab=readme-ov-file Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/

Read the full article here: www.marktechpost.com/2024/11/11/qwen-open-sources-the-powerful-diverse-and-practical-qwen2-5-coder-series-0-5b-1-5b-3b-7b-14b-32b/ Paper: arxiv.org/abs/2409.12186 Models on HF: huggingface.co/collections/Qwen/qwen25-coder-66eaa22e6f99801bf65b0c2f Demo: huggingface.co/spaces/Qwen/Qwen2.5-Coder-Artifacts Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/

Why Docker and not Podman?

Read our full take on Arcee-VyLinh : www.marktechpost.com/2024/11/07/arcee-ai-releases-arcee-vylinh-a-powerful-3b-vietnamese-small-language-model/ Model on Hugging Face: huggingface.co/arcee-ai/Arcee-VyLinh Details: blog.arcee.ai/introducing-arcee-vylinh-a-powerful-3b-parameter-vietnamese-language-model/ Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/

Read the full article here: www.marktechpost.com/2024/11/07/mbzuai-researchers-release-atlas-chat-2b-9b-and-27b-a-family-of-open-models-instruction-tuned-for-darija-moroccan-arabic/ Paper: arxiv.org/abs/2409.17912 Models on HuggingFace: huggingface.co/MBZUAI-Paris/Atlas-Chat-9B Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/

Read the full article here: www.marktechpost.com/2024/11/06/microsoft-researchers-introduce-magentic-one-a-modular-multi-agent-system-focused-on-enhancing-ai-adaptability-and-task-completion-across-benchmark-tests/ Paper: www.microsoft.com/en-us/research/uploads/prod/2024/11/Magentic-One.pdf GitHub Page: github.com/microsoft/autogen/tree/main/python/packages/autogen-magentic-one Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/

Read the full article here: www.marktechpost.com/2024/11/06/nvidia-ai-introduces-mm-embed-the-first-multimodal-retriever-achieving-sota-results-on-the-multimodal-m-beir-benchmark/ Paper: arxiv.org/abs/2411.02571 Model on Hugging Face: huggingface.co/nvidia/MM-Embed Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/

Great podcast. is this the real human voice or generated by notebookLM

It's a little hard to believe they're not reading from a script when they try to sound surprised by what the other one says, because they both already seem to know all about it. Kind of like the Howard Stern show.

omg , this podcast is so awesome , but its AI

how to build and setup this kind of ai stuf. i need step-by-step practicla instruction

Read the full article here: www.marktechpost.com/2024/11/03/meet-hertz-dev-an-open-source-8-5b-audio-model-for-real-time-conversational-ai-with-80ms-theoretical-and-120ms-real-world-latency-on-a-single-rtx-4090/ GitHub Page: github.com/Standard-Intelligence/hertz-dev Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/

All Watched Over by Machines of Loving Grace von Richard Brautigan I like to think (and the sooner the better!) of a cybernetic meadow where mammals and computers live together in mutually programming harmony like pure water touching clear sky. I like to think (right now, please!) of a cybernetic forest filled with pines and electronics where deer stroll peacefully past computers as if they were flowers with spinning blossoms. I like to think (it has to be!) of a cybernetic ecology where we are free of our labors and joined back to nature, returned to our mammal brothers and sisters, and all watched over by machines of loving grace

All Watched Over by Machines of Loving Grace von Richard Brautigan I like to think (and the sooner the better!) of a cybernetic meadow where mammals and computers live together in mutually programming harmony like pure water touching clear sky. I like to think (right now, please!) of a cybernetic forest filled with pines and electronics where deer stroll peacefully past computers as if they were flowers with spinning blossoms. I like to think (it has to be!) of a cybernetic ecology where we are free of our labors and joined back to nature, returned to our mammal brothers and sisters, and all watched over by machines of loving grace

Read the full article here: www.marktechpost.com/2024/11/02/meta-ai-releases-sparsh-the-first-general-purpose-encoder-for-vision-based-tactile-sensing/ Paper: ai.meta.com/research/publications/sparsh-self-supervised-touch-representations-for-vision-based-tactile-sensing/ GitHub Page: github.com/facebookresearch/sparsh Models on Hugging Face: huggingface.co/collections/facebook/sparsh-67167ce57566196a4526c328 Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/

Made with AI 😅

Ohhh AI presenters 😂

Read the full article here: www.marktechpost.com/2024/11/01/llama-3-nanda-10b-chat-a-10b-parameter-open-generative-large-language-model-for-hindi-with-cutting-edge-nlp-capabilities-and-optimized-tokenization/ Paper: github.com/mbzuai-nlp/Llama-3-Nanda-10B-Chat/blob/main/Llama-3-Nanda-10B-Chat-Paper.pdf Model on Hugging Face: huggingface.co/MBZUAI/Llama-3-Nanda-10B-Chat Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/

Read the full article here: www.marktechpost.com/2024/11/01/amd-open-sources-amd-olmo-a-fully-open-source-1b-language-model-series-that-is-trained-from-scratch-by-amd-on-amd-instinct-mi250-gpus/ Model on Hugging Face: huggingface.co/amd/AMD-OLMo-1B-SFT Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/

Can you eliminate all useless filler replies like "AHA" "OK" "EXACTLY" "GOT IT" ? It is really obnoxious.

Purely advertising no learning

Read the full article: www.marktechpost.com/2024/11/01/all-hands-ai-open-sources-openhands-codeact-2-1-a-new-software-development-agent-to-solve-over-50-of-real-github-issues-in-swe-bench/ GitHub: github.com/All-Hands-AI/OpenHands?tab=readme-ov-file#-how-to-contribute Details: www.all-hands.dev/blog/openhands-codeact-21-an-open-state-of-the-art-software-development-agent Audio Created by NotebookLLM and reviewed by real human 👉 Don’t Forget to join our 55k+ ML SubReddit: www.reddit.com/r/machinelearningnews/