Alibaba has launched its latest large language model, Qwen 2.5, which is designed to meet the growing demand for generative AI products. This new model, part of the Tongyi Qianwen series, is said to have significantly improved capabilities in reasoning, code comprehension, and textual understanding compared to its predecessor, Qwen 2.0. Alibaba Cloud claims that Qwen 2.5's performance is now on par with OpenAI's GPT-4 Turbo, marking a significant milestone in the AI industry.
WTF, @GroqInc . . . how is your #LPU inference engine speeding up over time; blazing 1,000+ T/s with @aiatmeta's colossal Llama-3-70b? This must be the fastest LLM for any stack, @sundeep @bensima. https://t.co/uzEgT5ttRj
Local LLM inference speed on CPUs is increasing! https://t.co/RDMvaA8rrY
llamafile is now the fastest way to run K quants on avx2. You should see prompt processing go 2x faster and text generation goes 1.3x faster than llama.cpp. Credit goes to Iwan Kawrakow for contributing his newest kernels. https://t.co/PTZdvrRAyu
Today in @AnythingLLM we made any LLM capable of being a reliable AI agent. Whether your LLM provider lets provides tool-calling or not. Here I take Llama3 running on @ollama and get a Perplexity level experience running all locally on my desktop. Real-time Web browsing and… https://t.co/8Os4t9jXp4
Zhipu AI continues to refine its models. In Jan 2024, it unveiled the GLM-4 model that claims to rival: - OpenAI's GPT-4 in many common benchmarks - Anthropic's Claude 2.1 in long text assessments - OpenAI's DALLE-3 in vision tasks Let's check the real numbers: https://t.co/KS8I8dXSZf
A new open source model with 10M context window is out. “We gradually increased context size from 32K →… 4M → 10M. This allowed us to prioritize pretraining with shorter sequences in the beginning, thereby offering higher utilization rates.” https://t.co/JK1lFiL0CZ https://t.co/7fh9kelOyl
Alibaba Cloud Unveils Enhanced Tongyi Qianwen 2.5 to Compete with GPT-4 Turbo #AI #AIinnovation #AItechnology #AlibabaCloud #artificialintelligence #ChatGPT #Chineselanguagefunctions #conversationlength #DingTalk #downloads #enterprisecustomers https://t.co/TsMnSWQLuZ https://t.co/OAyYbiLfXP
Alibaba $BABA unveiled its latest LLM, Qwen 2.5, amid rising demand for generative AI products, saying the model has "remarkable advancements in reasoning, code comprehension, and textual understanding" compared to Qwen 2.0. $BIDU $MSFT $NVDA
Rumors are swirling around new or potential new OpenAI releases 👀 The list includes: GPT-4.5 GPT-5 GPT-4.5 family (like Claude’s approach) GPT2 Search engine More GPT support Agents Agent platform Drastic improvement on reasoning (including explainability) Drastic…
Ok folks, it is time we hear LLMs coming from China (we have seen enough news from OAI, Google, Anthropic and meta). Now its Baba's time. $BABA Alibaba rolls out latest version of its large language model to meet robust AI demand. https://t.co/Hi7VAgWqS5
Alibaba Cloud officially released its latest version of the large language model Tongyi Qianwen(Qwen). Alibaba Cloud stated that the performance of this model has comprehensively caught up with OpenAI's GPT-4 Turbo. #Alibaba $BABA https://t.co/nEkFd1eIXT
Ok folks, it is time we hear LLMs coming from China (we have seen enough from OAI, Google, Anthropic and meta). Now its Baba's time. $BABA Alibaba rolls out latest version of its large language model to meet robust AI demand.
Alibaba rolls out latest version of its large language model to meet robust AI demand https://t.co/7nhEaUG5KB
A quick guide to using advanced AI models without limits: 1) Download 👋Jan: https://t.co/vv1CvwGYKA 2) Go to the model hub to get Llama3, Command R, or Mistral 3) Click "Use" to activate the model It runs locally & supports both llamacpp and TensorRT-LLM (for GPUs) engines ❤️ https://t.co/ty7LpE8s7Y
Reminder, it's been only 3 months since @GoogleAI released Gemini 1.5 w/1M context, 2 months since @AnthropicAI released Claude 3, beating GPT4, and 3 weeks since @AIatMeta released Llama 3, the top open source LLM by quite a ways. I think we're getting spoiled in 2024😁 https://t.co/SW6CJK4MPM
🐐 Llama-3 is the new hotness, but if you want to get the most out of the #model, then you need to #finetune. 👉 Check our #tutorial to learn how to easily fine-tune #Llama3 for customer service use cases. Notebook + data + $25 in free credits included! https://t.co/nMuNt459Ea https://t.co/aYbVs7VKdN
4M Context Length Llama-3 8B (V0.1) 🧑🍳 We’ve been actively pushing the boundary of long context models, which is why we’re excited to share an early version of our 4M context window Llama-3 8B on @huggingface. While perfecting a recipe takes time, this marks the longest… https://t.co/drxbSde7Aj
Let's talk about LLMs like LLAMA 3, by @AIatMeta. Cheap, efficient, and about to open a floodgate of possibilities. Imagine dozens (hundreds) of them working in cohort, ingesting 100K to 1 million tokens at a time, already made possible by cool people like @bindureddy and…