All tags
Topic: "pretraining"
nano-banana is Gemini‑2.5‑Flash‑Image, beating Flux Kontext by 170 Elo with SOTA Consistency, Editing, and Multi-Image Fusion
gemini-2.5-flash-image-preview hermes-4 nemotron-nano-9b-v2 internvl3.5 gpt-oss qwen3 deepseek-v3.1 google-deepmind nous-research nvidia openai ollama huggingface openrouter image-editing natural-language-processing multi-image-composition character-consistency reasoning hybrid-models context-windows model-steerability pretraining finetuning alignment vision vision-language api model-integration sundarpichai _philschmid lmarena_ai omarsar0 skirano yupp_ai xanderatallah officiallogank mervenoyann
Google DeepMind revealed Gemini-2.5-Flash-Image-Preview, a state-of-the-art image editing model excelling in character consistency, natural-language edits, and multi-image composition, dominating the Image Edit Arena with a ~170-180 Elo lead and over 2.5M votes. It is integrated into multiple platforms including Google AI Studio and third-party services. Nous Research released Hermes 4, an open-weight hybrid reasoning model focused on steerability and STEM benchmarks. NVIDIA launched Nemotron Nano 9B V2, a hybrid Mamba-Transformer with 128k context, top-performing under 10B parameters, and released a 6.6T-token pretraining subset. InternVL3.5 introduced 32 vision-language models based on OpenAI's gpt-oss and Qwen3 backbones. Ollama v0.11.7 added DeepSeek v3.1 support with hybrid thinking and Turbo mode preview.
Gemini 2.0 Flash GA, with new Flash Lite, 2.0 Pro, and Flash Thinking
gemini-2.0-flash gemini-2.0-flash-lite gemini-2.0-pro-experimental gemini-1.5-pro deepseek-r1 gpt-2 llama-3-1 google-deepmind hugging-face anthropic multimodality context-windows cost-efficiency pretraining fine-tuning reinforcement-learning transformer tokenization embeddings mixture-of-experts andrej-karpathy jayalammar maartengr andrewyng nearcyan
Google DeepMind officially launched Gemini 2.0 models including Flash, Flash-Lite, and Pro Experimental, with Gemini 2.0 Flash outperforming Gemini 1.5 Pro while being 12x cheaper and supporting multimodal input and a 1 million token context window. Andrej Karpathy released a 3h31m video deep dive into large language models, covering pretraining, fine-tuning, and reinforcement learning with examples like GPT-2 and Llama 3.1. A free course on Transformer architecture was introduced by Jay Alammar, Maarten Gr, and Andrew Ng, focusing on tokenizers, embeddings, and mixture-of-expert models. DeepSeek-R1 reached 1.2 million downloads on Hugging Face with a detailed 36-page technical report. Anthropic increased rewards to $10K and $20K for their jailbreak challenge, while BlueRaven extension was updated to hide Twitter metrics for unbiased engagement.
not much happened today
rstar-math o1-preview qwen2.5-plus qwen2.5-coder-32b-instruct phi-4 claude-3.5-sonnet openai anthropic alibaba microsoft cohere langchain weights-biases deepseek rakuten rbc amd johns-hopkins math process-reward-model mcts vision reasoning synthetic-data pretraining rag automation private-deployment multi-step-workflow open-source-dataset text-embeddings image-segmentation chain-of-thought multimodal-reasoning finetuning recursive-self-improvement collaborative-platforms ai-development partnerships cuda triton ai-efficiency ai-assisted-coding reach_vb rasbt akshaykagrawal arankomatsuzaki teortaxestex aidangomez andrewyng
rStar-Math surpasses OpenAI's o1-preview in math reasoning with 90.0% accuracy using a 7B LLM and MCTS with a Process Reward Model. Alibaba launches Qwen Chat featuring Qwen2.5-Plus and Qwen2.5-Coder-32B-Instruct models enhancing vision-language and reasoning. Microsoft releases Phi-4, trained on 40% synthetic data with improved pretraining. Cohere introduces North, a secure AI workspace integrating LLMs, RAG, and automation for private deployments. LangChain showcases a company research agent with multi-step workflows and open-source datasets. Transformers.js demos released for text embeddings and image segmentation in JavaScript. Research highlights include Meta Meta-CoT for enhanced chain-of-thought reasoning, DeepSeek V3 with recursive self-improvement, and collaborative AI development platforms. Industry partnerships include Rakuten with LangChain, North with RBC supporting 90,000 employees, and Agent Laboratory collaborating with AMD and Johns Hopkins. Technical discussions emphasize CUDA and Triton for AI efficiency and evolving AI-assisted coding stacks by Andrew Ng.
Apple Intelligence Beta + Segment Anything Model 2
llama-3-405b llama-3 segment-anything-model meta-ai-fair apple image-segmentation memory-attention video-processing pretraining cloud-tpus post-training synthetic-data instruction-following reasoning writing benchmarking bindureddy maximelabonne reach_vb
Meta advanced its open source AI with a sequel to the Segment Anything Model, enhancing image segmentation with memory attention for video applications using minimal data and compute. Apple Intelligence delayed its official release to iOS 18.1 in October but launched developer previews on MacOS Sequoia, iOS 18, and iPadOS 18, accompanied by a detailed 47-page paper revealing extensive pretraining on 6.3T tokens and use of Cloud TPUs rather than Apple Silicon. The paper highlights improvements in instruction following, reasoning, and writing through post-training and synthetic data. Benchmarks show Apple’s model scores lower than Llama 3, but with trusted human evaluations. Additionally, Meta released Llama 3.1 with a 405B parameter model, marking a significant open-source frontier model release.
Mistral Large 2 + RIP Mistral 7B, 8x7B, 8x22B
mistral-large-2 mistral-nemo-12b llama-3.1-8b llama-3.1-70b llama-3.1 llama-3-405b yi-34b-200k gpt-4o mistral-ai meta-ai-fair groq togethercompute code-generation math function-calling reasoning context-windows model-deprecation pretraining posttraining benchmarking
Mistral Large 2 introduces 123B parameters with Open Weights under a Research License, focusing on code generation, math performance, and a massive 128k context window, improving over Mistral Large 1's 32k context. It claims better function calling capabilities than GPT-4o and enhanced reasoning. Meanwhile, Meta officially released Llama-3.1 models including Llama-3.1-70B and Llama-3.1-8B with detailed pre-training and post-training insights. The Llama-3.1 8B model's 128k context performance was found underwhelming compared to Mistral Nemo and Yi 34B 200K. Mistral is deprecating older Apache open-source models, focusing on Large 2 and Mistral Nemo 12B. The news also highlights community discussions and benchmarking comparisons.
12/13/2023 SOLAR10.7B upstages Mistral7B?
solar-10.7b llama-2 mistral-7b phi-2 gpt-4 gemini upstage nous-research openai mistral-ai microsoft depth-up-scaling pretraining synthetic-data gpu-training api-usage model-integration agi asi chat-models vision model-performance fine-tuning
Upstage released the SOLAR-10.7B model, which uses a novel Depth Up-Scaling technique built on the llama-2 architecture and integrates mistral-7b weights, followed by continued pre-training. The Nous community finds it promising but not exceptional. Additionally, weights for the phi-2 base model were released, trained on 1.4 trillion tokens including synthetic texts created by GPT-3 and filtered by GPT-4, using 96 A100 GPUs over 14 days. On OpenAI's Discord, users discussed challenges with various GPT models, including incoherent outputs, API usage limitations, and issues with GPT-4 Vision API. Conversations also covered understanding AGI and ASI, concerns about OpenAI's partnership with Axel Springer, and pricing changes for GPT Plus. Discussions included the Gemini chat model integrated into Bard and comparisons with GPT-4 performance.